Advanced data analysis helping to nab Medicare cheaters

When FBI and Health and Human Services investigators announced earlier this summer they had snared more than 300 people trying to bilk Medicare out of $900 million, one might have concluded the corner had been turned in the fight against Medicare fraud. Fact is, it barely scratched the surface.

Last year, the Centers for Medicare and Medicaid Services (CMS), the federal agency that administers Medicare, estimated some $60 billion of American taxpayer money was lost to fraud, waste, abuse and improper payments. That’s more than 10 percent of Medicare’s total budget.

A major issue facing those who hunt down Medicare fraud is the sheer volume of claims and payments involved. Medicare’s contractors process 4.5 million claims every day.

The news of the 300 arrests in June marked a shift in the federal government’s approach to healthcare fraud. It’s a proactive strategy born out of the Affordable Care Act of 2010 – a $350 million boost to the Medicare Fraud Strike Force that marries investigative efforts with advanced data analytics.


Caryl Bryzmialkiewicz, the chief data officer and assistant inspector general at HHS, explained on Federal Drive with Tom Temin how the strike force uses data to find the culprits.

“We leverage CMS’s integrated data repository —about a petabyte of provider, beneficiary and claims data —to help us uncover suspicious patterns. We’re looking for trends in the data to help us understand fraud trends,” said Bryzmialkiewicz.

Even so, going through that much data looking for anomalies can be a needle-in-the-haystack affair.  Bryzmialkiewicz said that’s where teamwork comes in.

“We talk about using data analysis to support in one of two ways:  one is in a reactive sense, where we get a tip from our colleagues at CMS or from our Center for Program Integrity,” said Bryzmialkiewicz. She said tips also come from those  wanting to identify potential fraud  in their local community or from investigators with street intelligence about a particular fraud scheme. “If they see something, we can bounce that potential scheme against the data very quickly.”

The HHS data chief said a second way the data can be used by analysts is to look at the trends to try to predict future behavior. “We’re generating models to suggest we might want to have some additional scrutiny for certain providers for aberrant behavior,” she said.

One example was the case of a former Dallas doctor convicted in a nearly $400 million home health care scam. “(He) reported about 10,000 patients per month to home health.  The average is 100 patients per month,” said Bryzmialkiewicz. “He was definitely an outlier, so we were able to find that in the data very quickly.”

”Folks are really smart about hiding in the data, so we need  to figure out how to slice and dice it in unique ways to determine if  this might be somebody we want  to look at, we need to do that very quickly to find those potential candidates or provide additional evidence for cases already being pursued,” she said.

In fact, Bryzmialkiewicz said the data can be used in many ways against those who would cheat the government. It can be used by authorities to create an affidavit for a search warrant, for interviewing witnesses, in making a presentation to a grand jury, or to figure out size of a claims loss. Following a conviction, it can provide information to support a sentence.

“What I tell my team is ‘it’s kind of great if we’ve put ourselves out of business.’ So we’re also developing tools to help our investigators answer their own questions. If investigators are able to use the data while they’re working a case, that’s the exciting part of what we’re trying to do.”

The data analysts are also looking to expand the use of other techniques in the hunt for Medicare fraud, such as geospatial analysis. Much as the post office uses tools to sort the mail, the idea is to leverage the data in the hunt for fraud to the best extent possible. “A lot of the challenge in the data world is how you are integrating, how are you putting the information together. From my perspective, we also have law enforcement sensitive data, so we want to make sure we get truly actionable information to our investigators.”

Copyright © 2019 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.