How CMS takes on and beats back Medicare fraud

How CMS takes on and beats back Medicare fraud

In its second year of using the Fraud Prevention System, the Centers for Medicaid and Medicare Services identified or prevented $210 million in improper Medicare payments, according to CMS’s Report to Congress Fraud Prevention System Second Implementation Year, released in June.

That success can be attributed to two main aspects of FPS that set it apart from other fraud prevention systems, health payer analysts say.

“What makes this system a little bit different from many others, especially at this progression now, is that they try to identify the fraud before [payment] goes out,” said Ryan Blaney, a member of the Health Law group at law firm Cozen O’Connor’s Washington, D.C., office. “The other big difference is that it’s combining multiple systems working simultaneously.”

FPS uses predictive algorithms and other analytics to compare billing patterns against Medicare parts A and B fee-for-service claims prior to payment. The system is integrated with the Medicare claims processing system and other data sources, such as the Compromised Numbers Checklist of stolen provider identification numbers, the Fraud Investigation Database and complaints made through the Medicare call center. Another important resource is the Integrated Data Repository, which CMS stood up in 2006 as a storehouse of Medicare data, including beneficiary data, Part D drug information and seven years of parts A and B claims at three stages of processing.

Currently, FPS monitors fraud, waste and abuse via 74 models running simultaneously. In the past year, CMS added 39 models to the system, eight of which are predictive systems . Those are the most useful, Blaney said, because economists and engineers generate them by taking known cases of fraud and developing algorithms to identify claims that are more likely to be fraudulent than others.

“A single predictive model is often as effective as multiple non-predictive models,” the report states.

The other models fall into three other categories:

  • Rules-based, which searches for known stolen provider numbers, for example.
  • Anomaly, which searches for claims that lie beyond a baseline.
  • Network, which maps out associations that known offenders have.

When FPS detects a problem, it generates an alert. Then, CMS’ Zone Program Integrity Contractors begin investigating based on flagged providers that generate the most suspicion. Last year, CMS took administrative action against 938 providers and suppliers thanks to FPS, the report states.

The $210.7 million saved, almost double the amount identified during FPS’ first year in use, resulted in more than a $5 to $1 return on investment, up from last year’s $3 to $1 return, the report states.

The models come from CMS’s Analytics Lab, where teams of economists, statisticians and programmers develop and test them with the help of policy experts, clinicians, field investigators and data analysts.

CMS has used analytics to identify potential fraud before, but each model was run separately each month or quarter in a specific region of the country, the report states. In contrast, FPS runs the models simultaneously and continuously and with a national focus.

Commercial companies are also working to help the government detect and prevent fraud. For example, analytics company 21CT will launch a health care fraud detection solution early next month intended to help investigators find  leads, collect evidence and take action. IBM and CoreLogic also have fraud prevention solutions targeted to the government.

Moreover, fraud prevention isn’t for the health care sector alone. The Financial Fraud Enforcement Task Force, created in November 2009 by President Obama in response to the financial crisis, has StopFraud.gov, an online reporting tool broken down into categories such as mail fraud and cyber crime. The Justice Department’s Fraud Section also has a website for reporting illegal activity.

FPS sets a great example for other agencies and companies looking to thwart fraud, Blaney said.

“One of the advantages that health care has over some other agencies is a massive database of claims submissions,” he said.

“CMS has this robust database of information that they can take advantage of to run and help identify fraud. Taking advantage of this big data world and coordinating with investigators and the people that are knowledgeable about the various areas of your agency can identify fraud. I think that coordination and the use of big data can really go a long way, and it will in the future.”

Reader Comments

Tue, Sep 2, 2014 Jeff Leston Glen Carbon IL

all of these techniques are still "pay and chase" and require the adjudication of a claim to kick in. I don't see any mention of prevention, and the term "predictive" modeling is misleading; it is only a statistical indicator of a possibility, and requires more than one case to be substantiated most of the time.

Wed, Aug 13, 2014 Williams

This is great news , for far too long the healthcare industry has not gone in to check fraud that's been rampant ,  work with McGladrey and here's a piece on how data analytics and technology such as this will help in fraud detection I work for McGladrey and there's a very informative whitepaper on our website that readers of this article will be interested. @ “Data analytics is a powerful fraud prevention and policy enforcement tool” http://bit.ly/VSs1Mx

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above