Claims experts use data analytics to combat workers comp fraudReprints
Claims fraud continues to drive workers compensation costs up, driving payers to look for new ways to combat it.
Much like on the underwriting side of the insurance business, data and the use of technology to analyze it is seen by experts as a major development in improving efficiency and detecting problems.
The use of technology to sift through massive amounts of data to spot trends and anomalies will help claims professionals spot fraud more quickly, experts say.
Given the state-based workers comp system, it’s hard to get firm estimates on workers comp fraud nationally, but a figure often quoted for the whole property/ casualty sector is the National Insurance Crime Bureau’s estimate that it’s a $30 billion annual problem in the U.S. The NICB does not provide an estimate for the workers comp sector alone.
Widely reported examples of workers comp claims fraud include false claims, working while collecting benefits, payroll and employee misclassification, malingering injuries and medical fraud.
Part of the difficulty in assessing the level of workers comp claims fraud is the limited verifiable fraud data.
“In my personal practice I see exaggerations, I see malingering, I see a lot of suspicious activity. The amount of time when that rises to the level of actionable fraud is fairly limited,” said Chicago-based Rich Lenkov, attorney at Bryce Downey & Lenkov L.L.C.
However, the emergence of new technology is creating news ways to detect claims fraud. “There has always been fraud, but it’s becoming more easily detectable,” said Timothy Hopper, Stevens Point, Wisconsin-based special investigations unit major case manager at Sentry Insurance.
In 2016, close to 76% of insurers had integrated technology in their anti-fraud systems, with claims fraud detection leading, according to a 2016 study conducted by the Washington-based Coalition Against Insurance Fraud.
“When it comes to detecting fraud, technology is absolutely critical to the success of our program ... from a technology standpoint, data is the key — billing data, prior claim history data — so we can look at a situation where we may have a suspect claim involving a questionable medical provider. Our technology allows us to go out and look at other claims that match that same pattern. We can go out and look at other claims where that particular provider was involved, and we can compare billing patterns,” said Mr. Hopper.
And technological applications can detect fraud that might not be spotted by claims professionals.
“We use various tools like predictive models and analytic rules to try to find claims that might not have come to us from a reactive level but have certain elements that make it seem like there is a potential for fraud or have similar elements we have seen on prior fraud cases,” said Charlotte, North Carolina-based Eric Bushman, director of the commercial insurance special investigations unit at Liberty Mutual Insurance Co.
As a result, investigators can jump on cases based on the models rather than wait to be notified of suspicious claims, he said.
While technology has led to new ways to tackle the issue of detecting workers comp claims fraud, there are common red flags that can help identify fraud. Some of these red flags include the employee having a history of claims, no witnesses to the incident, the employee not reporting the injury or illness in a timely manner and the injury coinciding with a change in employment status, a Broadspire Services Inc. spokesman said in an emailed statement.
“Questionable or excessive medical treatment is a big red flag, a claimant that is hard to reach is suspicious, Monday morning reports of injury are always ones to scrutinize, and someone with a long history of claims inherently will be one that I look closely for suspicions of fraud,” said Mr. Lenkov.
Conflicting accident histories can also signal fraud, he said.
“It shouldn’t be complicated to tell your employer or medical providers how you got hurt if you legitimately got hurt,” Mr. Lenkov said.