The rapid growth in the use of data analytics in healthcare comes as no surprise to Eric Waller, senior vice president and chief marketing officer at Health Management Associates, a healthcare company based in Naples, Fla., operating 71 healthcare organizations across 15 states with $6 billion in annual net revenue.
Waller says when he joined the organization in 2009, he began looking for opportunities to apply analytics and quickly zeroed in on the revenue cycle as the best place to start.
"Analytics hold enormous potential on the clinical side, but the more immediate opportunities, at least for us, were on the financial side," Waller says.
"One natural place for us to start was in coding and billing," he says. "Historically, there have just been a lot of people thrown at the problem. Now, we've taken these experts and introduced math and machines, and we've seen significant results. We are making sure we are capturing the information and coding properly and getting the revenue we are due. We've developed sophisticated models to look for outliers in our coding. The machines crunch the data and identify the outliers."
Waller says that through the use of data analytics models, HMA is now preventing about $1 million per month in net revenue leakage.
"In a system of our size, if you make small, incremental improvements, the dollars add up quickly," Waller says. Also, HMA is achieving considerable savings by reducing its reliance on external resources. "We've eliminated a lot of the labor costs from outside firms that would normally have people doing manual sampling of files to look for missed codes. We've eliminated that expense, which was somewhere between $3 million to $5 million per year."
"Think about a room full of people just looking through files to make sure the coding is all correct," he adds. "That's not very scalable. Just throwing more people at the problem isn't the answer … We want to identify any areas where we could potentially underbill and also where we could potentially overbill. It's in everyone's best interest for it to not be over or under, but for it to be accurate. The most efficient way to do that is with computers and data analytics models."