Data Analytics and Medicaid Fraud: The Need for Standardization

Data analytics offers a proven and unparalleled opportunity to stop fraud, waste, and abuse (FWA), in state-run Medicaid programs. However, the lack of standards and federal leadership has created a very costly stumbling block.

According to numbers provided by the Centers for Medicare and Medicaid Services (CMS) in 2015, improper payments account for about 5 cents of every Medicaid dollar. This equates to approximately $29.1 billion of the $547.7 billion program. Those improper payments fall into three categories:

  1. Blatant Fraud: Fraud by deception or misrepresentation
  2. Accidental Misrepresentation: The naïve use of services and resources, such as the use of incorrect billing codes.
  3. Unethical Practices: Provider practices that are inconsistent with ethical medical and business practices, such as providing medically unnecessary services, or beneficiary practices. In other words, prescribing medication or performing a procedure a patient does not need.

Without automation, agencies fight a losing battle

Identifying these categories remains challenging.  State and Federal law enforcement agencies can data mine and follow up on a suspected lead, but the sheer volume of claims is so great. Advanced data analytics and predictive analytics promises to improve detection, yet without consistent, standardized practices and metrics, and sufficient governance, it leaves many states to fight these battles alone.

On the Federal level, CMS demonstrates success in fighting FWA in Medicare. Medicaid programs however, are state run, each managed independently by state and each with their own datasets, metrics, and formats.

CMS uses its own Fraud Prevention System (FPS) to detect Medicare fraud, like those used by banks and credit card companies to identify credit and debit card fraud. By applying analytics to claims and invoices at inception, CMS can identify suspicious billing patterns. Since implementing FPS in 2011,

CMS has generated $820 million in savings – a 10:1 return on investment in the program’s first three years alone.

Deep dive into data will help identify patterns and anomalies

For states, most FWA programs rely heavily on people, rather than automated advanced analytics. The typical process relies on a manual process in which it applies rules and algorithms to review checks, detect claims anomalies, and spot patterns. Painstakingly, employees can discover FWA through a combination of cost avoidance and cost recovery. While strong analytics are a key factor in FWA detection, it highlights the importance of prevention through predictive analytics.

Trade-off: Pay less in a shorter window

These changes imply that Medicaid and Medicare charge lower rates to participating practitioners, but in exchange must pay their participants promptly and consistently. This fact alone screams for automation in pre-payment analytic work. And yet, most states are still ill-prepared. Several states don’t have the bandwidth to create specialized FWA teams, let alone the ability to effectively automate.

Stumbling Blocks

Because of the complications around manual analytics and recovery, most states will have stumbling blocks to success. Here are a few to keep in mind:

  • Without full data access, employees cannot uncover critical information, such as cases in which investigators find fraud at one provider and later track its ownership to other providers accused of similar crimes.
  • Patient privacy protections limit regulators.
  • Legacy processing systems that use proprietary data formats have cumbersome interfaces. These make integration both a structural problem and a technology challenge.

The solution to these stumbling blocks is simple: Create a centralized national database.

Federal and state governments need to better develop guidelines for data sharing and governance. The standardization of forms, form fields, metrics as well a national database and secure method for data transmission will aid in creating consistency across state lines. This would include payment information, relationships between people, ownership connections, real estate records and other unstructured data, to also incorporate the use of metadata and socioeconomic data. As a result, it will make it more difficult for fraudsters to escape detection by resettling their operation to another part of the country.

By giving state and federal stakeholders a chance to discover more accurate algorithms and trends, they can better identify outliers and develop a common understanding of fraud indicators within a given program. 

 

Learn more about Archetype's Healthcare Analytics pratice by clicking the button below.

 Learn More