High complexity claims account for the majority of dollars spent in workers’ compensation claims. One way to identify high exposure claims early is by using predictive analytics. Using analytics, carriers can spot problem cases and direct them to a more appropriate outcome.
Nearly 83 percent of personal lines insurers and 85 percent of commercial lines insurers planned increases in spending for data and analytics between 2014-2016, according to Verisk Analytics figures.
“Today, the number one area of focus is in claims,” said Adam Wesson, director of claims solutions for ISO Claims Partners, a Verisk Analytics company. He and Rob Lewis, the president of ISO Claims Partners, spoke on the topic recently during Verisk’s Monday webinar series.
The top five areas being invested in:
Key market trends influencing business intelligence, according to Wesson, include enterprise data, technology, business and talent. Insurers want new analytics in order to minimize operational risk, leverage emerging technologies, adapt to consumer desires and for industry benchmarking purposes.
According to Lewis and Wesson, identifying high exposure claims early allows insurers an opportunity to change the outcome. Using objective data is helpful to improve decision making on a claim, Wesson said.
Lewis explained that 80 percent of claim costs are driven by five percent of workers’ comp claims. The webinar, Predictive Analytics: Putting the Brakes on Out of Control Workers’ Compensation Claims, highlighted some of the reasons why claims go bad, including:
- Late reporting;
- Poor initial diagnosis;
- Doctor hopping;
- Lack of modified work duty options;
- Adjuster inexperience;
- Lack of manager oversight;
- Lack of proactive case management;
Large loss claims can be unpredictable, said Lewis, that’s why it’s critical for claims departments to use predictive analytics. In fact, large loss claims may begin with what might be considered small injuries, according to The Hartford’s Kevin Finn, vice president, National Accounts, Captives and Specialty Programs. In the insurer’s Perspective article on the subject, he said that continuous monitoring is key. The Hartford’s predictive models, which are run on every claim, begin at the first notice of loss and continue to run throughout the claim. Verisk’s predictive model, WC Navigator, can identify claims severity early and offer early reassessments for continuous decision-making.
In an Advisen white paper sponsored by ACE and released last year, Keith Higdon, senior vice president of client services at ESIS, pointed out that sometimes claims aren’t severe at the outset but that a predictive model can alert an adjuster that similar claims took a turn at a certain period in the claim cycle.
“Predictive analytics can help to make better decisions earlier and improve outcomes throughout the life of a claim,” said Lewis.
On a wider scale, predictive modeling can minimize operational risk by reducing leakage and cycle times as well as assist in situations where seasoned staff have retired.
Sources of data that can be used in predictive analytics include health, disability and workers’ comp claims data.
According to the Advisen white paper, data can be structured – like information found in formatted fields i.e., claimant demographics. Data can also be unstructured, such as information found within an adjuster’s case notes. The paper also noted that there is abundant data at the claim level to produce a better outcome for both employers and employees.
In the absence of predictive modeling, adjusters may not understand why a claim is still open after several years.
The Power of Predictive Models in Workers’ Compensation Claims Handling authors wrote, “with huge volumes of crucial data being held in an unstructured format and experience not captured by formal claims systems, many triggers that could provide an early alert to potentially large claims, can easily go un-noticed even when the claims handler is diligently managing a claim in the normal course of business.”
Even with analytics, claims still require analysis by an adjuster.
“Predictive models are tools for adjusters and team leaders and can only work if combined with their experience and judgment,” said Higdon.