Postal Service Busted Up a Billion-Dollar Drug Problem with Data


A scam attempting to fleece the government for millions of dollars was unearthed thanks to data analytics.

Editor's note: This article was updated to correct the decrease in Postal Service spending on compound drugs in fiscal 2017.

About two years ago, analysts within the U.S. Postal Service’s Office of the Inspector General noticed a spike in health care costs revolving around workers’ compensation claims and spending on prescription drugs.

As the office’s data gurus used analytics tools and algorithms to dig into the data, they realized the increased spending was a coordinated effort between several parties attempting to fleece the government for millions of dollars. The scam focused on compound drugs, which allows multiple medications to be blended into a single pill but is billed full price for each medication.

During the nine-month period from January to September 2015, the daily cost of compound drugs for the U.S. Postal Service grew from an average of $250,000 per day to more than $390,000 daily. The agency spent almost $180 million on them in fiscal 2016. In 2013, it spent less than $10 million on the same compound drugs, reimbursing the Labor Department.

» Get the best federal technology news and ideas delivered right to your inbox. Sign up here.

Traditionally, Labor covers workers’ compensation claims and seeks reimbursement from individual agencies through invoices.

However, as Kelly Tshibaka, chief data officer within the Postal Service’s OIG, explained last week at an event hosted by Nextgov, Labor wasn’t doing oversight on payments going out. Her team uncovered “a conspiracy in a couple states” between providers and compound pharmacy owners.

In California, for example, doctors were paid to prescribe compound drugs patients didn’t need, according to the Postal Service OIG’s lengthy audit. Sometimes, Tshibaka said, providers owned compound pharmacies—a violation of federal law—directing recipients to compound pharmacies, overbilling Labor for their medications and then splitting profits.

Tshibaka said her team turned the data over to investigators and agents who created successful cases. Importantly, however, the data better informed Labor of how vulnerable it was to fraud.

As of last year, Postal Service employees must provide a medical justification from a doctor in order to receive medication from a compound pharmacy. In fiscal 2017, Postal Service expenditures on compound medications decreased some 78 percent.

Tshibaka said the Postal Service OIG’s investigative efforts led to estimated long-term savings for the agency of $1.2 billion in health care spending, and that’s only one of its many cases.

In total, the Office of the Chief Data Officer she heads has used analytics to peruse millions of pages of paperwork to produce more than 500 leads for investigators across the $13 billion the Postal Service spends on contracts annually, with “only one of them a false positive.”

“It’s one of our biggest signs of success,” Tshibaka said. “It means every lead our agents are working with is worth their time.”

The office includes analysts and data scientists, and has buy-in from the executive levels of the Postal Service because of its track record. They are skilled in predictive and geospatial analytics, machine learning, regression analysis, sentiment analysis and creating advanced visualizations that “tell their stories” to decision-makers and investigators. Tshibaka said the team prioritizes potential cases based on factors that generally lead to successful prosecutions, such as cost overruns or scope creep, to ensure the best use of their time and that of investigators who would work their cases.

The entire OIG office, which consists of 1,100 people and a dwindling budget, still provides exceptional oversight to the Postal Service and its 600,000-strong workforce, she said.

“Last year, our tools and models helped us contribute to $920 million in findings,” Tshibaka said. “Anyone in the position of having to do more with less? Your answer is data analytics.”