Predictive analytics will, for the first time, help the Center for Medicare & Medicaid Services (CMS) meet its goal to prevent fraud via its Fraud Prevention System (FPS), but the program suffers from inconsistent data and flawed methodology that make it impossible to track inaccuracies, according to the Office of the Inspector General (OIG), under the U.S. Department of Health & Human Services.

In July 2011, CMS launched its new FPS, which was designed to use predictive modeling and data analytics to review all Medicare fee-for-service claims for indications of fraud. CMS' relative inexperience with predictive modeling is clear, according to a Dec. 17 OIG report. To improve results, the OIG recommended CMS:

• Require contractors to track money recovered from following FPS leads;
• Coordinate with law enforcement to better report outcomes of cases referred from FPS;
• Revise its methodology for calculating projected savings from improper payments to recognize that not all claims from a revoked provider are necessarily false and that some previously denied claims may eventually be paid; and
• Revise the methodology for calculating costs avoided to verify certain information and to include costs associated with the FPS.

CMS agreed to improve its savings methodology in a report to Congress which also said the agency already had:

• Improved how it tracked recoveries of overpayment and better estimated law-enforcement recoveries;
• Found ways to minimize or eliminate deficiencies in manual data reporting;
• Evaluated a possible fix for systematically accounting for legitimate services and claims overturned on appeal; and
• Incorporated costs identified by OIG in calculating return on investment.

CMS noted that the FPS "generated leads for 536 new fraud investigations, provided new information for 511 pre-existing investigations, and triggered thousands of provider and beneficiary interviews to verify legitimate items and services were provided to beneficiaries."

CMS said it is the first federal agency to use predictive analytic technology on a large-scale purpose to identify fraud, waste and abuse.

CMS came under fire in November by the Government Accountability Office (GAO) for being slow to develop the predictive-modeling system, including integrating with its payment-processing system. The GAO also said CMS had not yet defined or measured quantifiable benefits of performance goals.