The Centers for Medicare & Medicaid Services is shifting away from its inefficient "pay and chase" approach to recovering payments for fraudulent claims by implementing measures to prevent those claims from being paid in the first place. Beginning July 1, 2011, all Medicare fee-for-service claims will be analyzed with predictive modeling software. Providers need to examine their claims practices and compliance programs to ensure adequate support for claims that may be identified as suspect by the software.

Predictive Modeling of Medicare Claims

The Centers for Medicare & Medicaid Services (CMS) is introducing predictive modeling technology as part of the Obama Administration's effort to reduce the estimated $60 billion that government health care programs lose each year due to fraud and abuse. The CMS predictive modeling program is authorized by Section 4241 of the Small Business Jobs Act of 2010 (the "Act"). The software will evaluate all incoming Medicare fee-for-service (FFS) claims for factors suggesting potential fraud or abuse. The Act allows CMS to avoid Medicare's statutory prompt payment requirement when blocking reimbursement of claims flagged by the system for further review.

Private health insurance companies, banks, and credit card issuers already use similar software to detect fraud in their claims systems. Analysts expect that CMS' predictive modeling software will share some characteristics of existing private sector applications. The CMS fraud-detection model may incorporate the following factors already in use by the software of commercial health insurance companies:

  • Improper "unbundling" of services that should be billed as one procedure, but instead are charged as multiple, individual procedures.
  • "Upcoding," or utilizing reimbursement codes for services with higher reimbursement rates than the rates of the services actually rendered.
  • Claiming a high percentage of provider workdays that exceed eight hours in length.
  • Claiming a recent increase in weekends worked by providers.
  • Increase in service lines billed across multiple patients and multiple claims.
  • Increase in the use of claim line modifiers.1

Investigators will review medical records and other information associated with the records that are flagged as potentially fraudulent by the software. Legitimate claims are to be paid promptly. Claims requiring further review may be referred to the Health Care Fraud Prevention and Enforcement Action Team, a joint venture between the Department of Health and Human Services and the Department of Justice dedicated to prosecuting health care fraud.

Commercial health insurance companies have used predictive modeling as an efficient and effective way to combat fraud. The nation's largest private health insurer, UnitedHealthcare, saved $125 million over two years by using an analytics system similar to the one that CMS will debut in July. Kaiser Permanente has also used predictive modeling software to identify suspect providers. This success of private sector appli-cations makes CMS optimistic regarding the technology's potential to reduce fraud and abuse in Medicare.

Expansion to Other Programs

The Act requires CMS to expand the predictive modeling program to Medicaid and the Children's Health Insurance Program ("CHIP") by April 1, 2015. This expansion is contingent on the software's performance in the Medicare program and an analysis of the cost-effectiveness of monitoring insurance programs that are partially administered by the states. However, the speed of the technology's implementation in Medicare suggests that CMS may begin using the software in Medicaid and CHIP ahead of schedule.

Relationship to Other Anti-Fraud Initiatives

Predictive modeling technology will complement several recent anti-fraud initiatives. HHS recently announced that a portion of the $350 million in anti-fraud funding provided by the Patient Protection and Affordable Care Act of 2010 (the "ACA" or "PPACA") will be used to hire law enforcement agents to pursue health care fraudsters. In conjunction with the Veterans Administration, the Department of Defense, the Social Security Disability Insurance program, and the Indian Health Service, HHS has begun pooling its claims data with information from other government agencies to help improve fraud detection. The CMS predictive modeling program will benefit from the enhanced enforcement resources and access to data created by these measures.

Conclusion

CMS' adoption of predictive modeling technology has significant implications for health care providers. Proper recordkeeping and regulatory compliance are essential to provide support for claims that may be flagged for investigation by the software. Providers should be prepared for delays in payment as a result of these investigations, and to justify claims or resubmit claims in their entirety. To avoid scrutiny, physicians and other providers should examine their claims practices and determine the risks for audits triggered by the predictive modeling software. Providers should seek guidance on compliance activities and efforts to implement acceptable billing procedures.