Using Segmentation to Drive Efficiency in Healthcare Collections

Jason Petrasich

When people owe money for healthcare services they have received, we know 90% of them will either pay their entire bill or none of it at all. The challenge, and therefore the key to success, is figuring out which person is which so you know where to spend your time in collections.

Successful healthcare collectors use a process called segmentation to predict how likely a person is to pay the bill they owe for care received at a doctor’s office or hospital. One company recognized as a pioneer in the use of regression analytics and population segmentation in the revenue cycle industry is Conifer Health Solutions, which is headquartered in Frisco, Texas, and began this process back in 2004. Over the past seven years, Conifer has continued to build and improve on the process, and now uses a process called MicroSegmentation™ to combine data modeling with its day-to-day operations.

“It’s an unfortunate truth in healthcare that a large percentage of patient bills go unpaid, especially where providers carry the burden of a large uninsured population,” said Conifer Senior Vice President of Operations Jeffrey Nieman. “Segmentation allows healthcare providers (and collectors) to identify the population that is very unlikely to pay, regardless of how much work effort is applied. Hospital CFOs appreciate the need for lighter work effort on these accounts, as well as the reduced expense of minimizing efforts on that population. The resources that are freed up can then be redirected toward accounts that have a much higher likelihood of payment.”

Nieman stated their model accurately predicts the 30 percent that will most likely pay their bill and the 30 percent that won’t, leaving the remaining 40 percent of accounts as the primary focus for the collection staff. He cautioned, though, that while knowing which accounts are likely to pay or not is important, it doesn’t necessarily translate to revenue cycle efficiency without also having an ability to act on that information.

“For example, take patients that were treated in the ER vs. those that had an elective service. We have found a significant difference in each group’s willingness to pay regardless of their ability to pay. So we built our model to drive customized workflow strategies to leverage that information. Different segments get different numbers of letters, calls and other work efforts, and even the timing of those efforts is fine-tuned for each population to maximize the response rate,” Nieman said.

Nieman noted that building a successful segmentation model requires more than considering credit score alone. He said the industry’s more advanced models bring together dozens of data points, recognize how that data has driven behavior historically, and even enable new accounts that come into the process to be recognized as to which segment of the population they are likely to fall in.

Jason Petrasich is the Senior Director, Applied Research, Revenue Cycle Solutions for Conifer Health Solutions.