Predictive Analytics and Medicaid Program Integrity

optum pi_FotorPublic benefits programs such as Medicaid are multidimensional, complex and continuously evolving. Because of concerns about program size, growth, diversity and adequacy of fiscal oversight, the Government Accountability Office (GAO) first designated Medicaid at high risk for fraud, waste and abuse in 2003, and it remains a high-risk program 12 years later. In fiscal year 2014, Medicaid distributed $508 billion, of which state governments shouldered $204 billion. With an overall improper payment rate of 6.7 percent, Medicaid lost more than $34 billion — and states bore the burden for about half that amount.

Multiple state and federal organizations participate in Medicaid program fiscal oversight, but most of the burden falls to the states, which have the primary responsibility for funding and administering program integrity (PI) teams to minimize tax dollars lost to fraud, waste and abuse. Historically, state PI initiatives have focused on retrospective recovery of paid claims — the “pay and chase” effort to recover improper payments after they have been made. Overpayments have the potential to increase as Medicaid becomes larger and more complex. It’s common for providers to make coding mistakes or misunderstand ordering and billing rules, and criminals are growing more sophisticated. States must take their PI game to the next level by integrating cost avoidance activities — based on predictive analytics — into their fraud, waste and abuse strategies.

This handbook serves as a starting point for states seeking to re-invent and re-invigorate their Medicaid PI initiatives and processes with predictive analytics approaches and tools. It recommends a more integrated and comprehensive approach, including:

  • Adding predictive analytics into the front end of the claims cycle to avoid overpayments
  • Developing a comprehensive, analytics-based approach to screening and eliminating problem providers
  • Supplementing conventional post-payment recovery efforts with predictive analytics


Leave a comment

Your email address will not be published. Required fields are marked *