Claim triage is hardly a new concept. Every insurer looks to achieve early identification of those workers' comp claims that run the risk of high medical costs and a long payout period. Early intervention into those cases could give insurers more control over the handling of the claim, the medical treatment, and the overall costs with an earlier return-to-work.
Serious injuries that result in the loss of a limb or severe burns are obvious candidates for claim triaging, but, what about those developmental claims that are not as obvious? On the surface, developmental claims may appear trivial, and may not raise any red flags. As time goes on, however, these claims start to show their true colors in the form of exorbitant managed care costs and out of control medical expenses.
While insurers may have the processes in place to identify the 'usual suspects', how can they identify those that are not as obvious to take advantage of early intervention while controlling costs and getting workers back to work sooner rather than later?
Sobering Statistics
There is a seemingly endless supply of statistics that shape insurance company strategies as they attempt to mitigate risk, and best manage exposures while keeping their eyes on generating profit. Companies selling workers' compensation insurance are no exception, and they're facing some sobering industry statistics that are prompting many insurers to take a closer look at their claim management strategies.
According to the NCCI's winter 2009 research brief, the duration and severity of time loss claims has continued to rise with average indemnity up 4 percent between 2002 and 2008, while the actual number of time loss claims also continues to increase. Wage inflation during the same period increased less than 4 percent, while average medical costs rose over 6 percent.
NCCI research also revealed that about 75 percent of the total number of claims is in the $1,000 to $50,000 claim size range, while 80 percent of the ultimate dollar costs associated to claims is in the $10,000 to $500,000 range. Without even looking at the makeup of medical services associated with large cost claims, while including the fact that according to the Medical Consumer Price Index (CPI) the average prices of medical services increased 45 percent between 1998 and 2007, it becomes clear that the smaller number of large loss claims are responsible for the majority of incurred medical and indemnity costs.
The average national combined ratio for workers' compensation is now approaching 120 with no discernable timeframe for a turnaround. This number combined with the NCCI's claim statistics provide the backdrop for very challenging times for workers' compensation insurers.
What's An Insurer To Do?
Many large loss claims are easy to identify and can be subjected to early claim triaging protocols designed to minimize costs while maximizing the probability of getting a claimant back into the workforce quickly. Yet, developmental claims where cost is not so clear-cut remain. These claims are at the root of many of the exposure problems that insurers are facing today, primarily because success in managing claims effectively is directly related to the timeframe in which treatment protocols are enacted. For developmental claims, the opportunity for early claim triaging and intervention is often lost in the guise of claims that look fairly harmless.
Insurers have implemented a variety of techniques to identify those developmental claims in an effort to manage their exposures. Approaches range from the analysis of standard data, to the use of predetermined decision rules when certain characteristics exist, to quasi-profiling of all claims in an attempt to provide quality treatment for claimants as early as possible. Unfortunately, despite all of the valiant efforts, attempts to identify developmental claims earlier rather than later have been met with only modest levels of success, while large cost development continues to persist and sometimes worsen.
One technique that is beginning to attract some attention from insurers is the use of predictive analytics. While certainly not a new concept, predictive analytics has been used to support reserving or to identify potential fraud for years. Now insurers are beginning to explore how predictive analytics can be used to analyze the historical claim data, uncover hidden claim characteristics, and identify trends that will allow them to triage developmental claims that may not currently be flagged for early intervention. When used effectively, predictive analytics can have a dramatic impact on the avoidance of medical costs and indemnity associated with developmental claims.
The Need for Data
While it may be a viable approach with significant potential value, utilizing predictive analytics presents insurers with a unique set of challenges. Those insurers trying to identify developmental claims through the use of predictive analytics need to have the ability to compile the data necessary to build a good predictive model. For predictive models to work effectively, the data needs to not only provide a good deal of depth and volume, but it also needs to span many characteristics, not all of which are easily discernible or extractable for analysis. While the insurance industry's custodians of data typically have a large quantity of data, the data sets are often too narrow in scope. Consequently, the models are often unable to produce the desired results.
There are ways that insurers can address this data challenge by bringing additional claim data to the table.That additional data may include not only bill review data and diagnostic data, but also demographic, econometric, firmographic, and other less obvious data sources that might require a combination of third-party data partners in order to build the desired predictive models. Presuming that the quality and quantity of data is acceptable, current, and is being provided in a way that supports refreshing the model for continued and accurate predicative capabilities, predictive analytics could spell the difference between success and failure in the early identification of developmental claims.
Getting Started
While predictive analytics can offer new and unmatched insight into the business which may translate into significant value, implementation of these complex technologies is not easy. An insurer should do an honest self-assessment of the in-house predictive analytics skill set, and seek help from experts when necessary. To increase the likelihood of success, insurers would be well-served to engage the services of an experienced modeling group that has experience with insurance as well as the proven disciplines and best practices required to build and validate the desired predictive model.
Expert qualifications required include, but are not limited to, demonstrated mastery of disciplines associated with the segmentation of data; the identification of predictive variables; and how to validate models through blind tests conducted on well-established holdback samples. The modeling group must be able to work closely with the insurer to transfer knowledge during the critical process of further refining the target that the model will address and upon which the effectiveness of the model will be based. The joint team effort will also allow both parties to establish the definition of desired data, the identification of the data sources, and the availability of the data to support the effort.
Presuming that minimum requirements are met and there is a team in place that has the skills and the tools to get the job done, the effort of building and testing the model can begin in earnest. Once completed and proven, the model will need to be deployed, continually monitored, and tuned to insure that the predictive capabilities remain high. Insurers need to view predictive analytics as an ongoing journey rather than a one-time destination, or the initial model will quickly become outdated and lose its value.
Insurers who successfully gain access to or build and maintain these models will be provided the opportunity to intervene much earlier in the claim process, directly impacting the level of service provided to claimants and resulting claim costs. When properly implemented and maintained, the benefits of applying predictive analytics to identify developmental claims earlier improve over time as the quality, breadth, and depth of data is improved. When developmental claims are flagged early in the process, not only are medical and indemnity costs reduced, but the injured worker is able to receive the best support and reenter the workforce as soon as possible.
Predictive analytics coupled with other actions, such as rule-based decision making, profiling, and text mining, will provide claim triaging best practices, directly contributing to an insurer's ability to control more of the risk their business faces. As claim costs continue their upward trend, insurers should also ramp up their efforts to find new and innovative ways to reduce costs on all fronts. With properly implemented predictive analytics, insurers now have a fighting chance to gain ground in this ongoing battle.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.