Fraud prevention: Decipher the claims data story


Payers sift through and analyze millions of provider claims, working to prevent fraud and cut costs. They rightly consider these claims as valuable points of information that help identify high-quality doctors, improve business processes and streamline workflows.

But what if instead of viewing claims merely as data files, insurers considered claims as a story, replete with a developing plot line that exposes the protagonists and antagonists? With that outlook, insurers could read the claims like a book, uncovering the motives of each character and decoding the story's plot line more easily.

And after reading through so many similar stories, insurers likely would catch on and begin predicting the characters' next moves--and even blocking them from taking unwanted action, such as fraud.

"The claims tell a story. The data tell a story. You just have to decipher it," Alanna Lavelle, director of investigations at WellPoint, Inc., told FierceHealthPayer for its latest eBook, "Payers Embrace Big Data."

Key to deciphering claims is structuring the data like a reader might chart pages of a complex narrative in a book. "Claims data must be harmonized, standardized and brought up into a groomed layer of information" before payers can use them for purposes beyond reimbursement, said Pamela Peele, Ph.D., chief analytics officer at the University of Pittsburgh Medical Center Health Plan.

Once the claims data are appropriately structured, payers can mine the information for red flags or abnormalities. In one case, Blue Cross and Blue Shield of Louisiana analyzed data for a 200,000-employee group and uncovered a chiropractor's office was seeing staff at a rate and duration far above specialty norms. An on-site review then showed patients received repetitive, massage-like services that didn't qualify for payment. BCBSLA assessed an overpayment of several hundred thousand dollars.

As payers look to become more effective at detecting potential fraudulent claims, they're using predictive analytics by building sophisticated models that are continuously refined based on their own data. "Predictive modeling creates a positive feedback loop where the data get better and the fraud prevention models more sophisticated," Louis Saccoccio, CEO of the National Health Care Anti-Fraud Association (NHCAA) told FierceHealthPayer.

In fact, the NHCAA believes so strongly in these predictive approaches, that it advises all payers "to get on board ... to make a meaningful dent in the fraud problem," Saccoccio added. "What you learn post-payment today you might be able to use prepayment in the future," he said.

Although predictive analysis is still in the early stages, the general best approach is for payers to focus on how they can gather additional types of data earlier in the process, said Mark Golberg, general manager of the provider, payer and ACO sectors at consultancy Recombinant by Deloitte.

Payers could, for example, increase their claims pattern recognition by using patient risk assessments, patient satisfaction surveys and telemetry. "Watch for things like a patient receiving diabetic medications and supplies but not providing glucose data to their provider," Golberg explained. He added that payers should take advantage of certain social media sites like StreetRx and Twitter for pricing and doctor information. - Dina (@HealthPayer)

Editor's Note: Interested in learning more techniques to harness big data in your organization? To download the free FierceHealthPayer eBook, "Payers Embrace Big Data," click here.