ICD-10: Providers can recoup millions of dollars in lost revenue by analyzing claims denials, data scientist says

Advanced analytics and machine learning technologies are critical to pinpointing problems in large datasets that could be losing providers money. That’s why some organizations are investigating every single denied claim to better understand trends. 
By John Andrews
10:12 AM

If claims are the lifeblood of providers' operations, then denials are a virus that threatens their financial health. That means understanding the reason why claims get rejected, especially with the ultra-granularity of ICD-10, is paramount.

Tracy Dean, business office director for OrthoTexas Physicians and Surgeons, said that since the transition her organization does not assume any denial is a one-time, isolated event.

"We are working the claims and looking at denials ― nothing is too small not to have a conversation with the team," Dean said.

Indeed, many providers are facing the need to adjust to correctly report the data, according to Allison Gilmore, principal data scientist for healthcare with Menlo Park, California-based Ayasdi.

Ayasdi develops advanced analytics and machine learning software for claims-denial management. Gilmore said its apps can extract denial and claim trends from extraordinarily large datasets to recoup potentially millions of dollars in lost revenue.

This article is part of a Special Report on ICD-10. Other stories include a look at ICD-10 at 6 months: Providers and payers have yet to reach the bottom, and how the new code set is poised to usher in a new generation of younger, more tech-savvy medical coders. 

"Denials are important. The data is complex," Gilmore said. "Denials are not always the same, so when trying to understand denial patterns, it's a small percentage of claims and the trends are smaller than that."

ICD-10 presents a number of challenges, but they can be distilled down to two prominent ones, she said: the speed with which new trends arise and a negative cause-and-effect pattern within code analysis.

In addressing the first challenge, Gilmore noted, "new trends come up fast and to identify them you only have a couple months of data to review so far ― the data pool is small."

On the second point she observed: "When you expand out the number of codes, you decrease the number of claims with any given code. We call that sparsity and ICD-10 just made things harder."

So how complicated has ICD-10 made operations?

"From a clearinghouse perspective, HIPAA 5010 was a bigger issue," said Matthew Ketterman, director of product and portfolio analysis at Florida-based Availity.

Provider claims, in fact, have been clean since the deadline, Ketterman said, and claims payments have been consistent, with a less than 1 percent rejection rate ― which makes understanding those denails all the more critical. 

At OrthoTexas Physicians and Surgeons, in fact, every single instance is important.

"Each denial,” Dean said, “could be part of a larger pattern."


Twitter: @HealthITNews

Like Healthcare IT News on Facebook and LinkedIn