This article was first published on Claims Journal.

Ask claims adjusters what they think of Medicare Set-Asides (MSAs) and you’ll likely get a wide range of answers. While most would agree, in theory, that MSAs serve a valid purpose, the way they are constructed today is labor-intensive, slow, and often requires a significant number of emails between the adjuster and the allocator to arrive at a final MSA fine-tuned to the needs of the claim.

But what if there was a way to modernize MSAs with the help of technology? That time has come. The opportunity exists for MSAs to be generated in record time, with the same degree of accuracy, by putting the adjuster in the driver’s seat and allowing them to specify coverage under the claim when requesting the MSA.

To fully understand what the future of MSAs looks like, let’s first examine why they need to be pulled out of the dark ages to begin with.

MSAs Today

MSAs are created by examining an injured worker’s medical records. But before an allocator can go through those records, they have to obtain them, which can take a long time. It often involves someone physically making copies of 400 to 500 pages of records from a hard copy claim file or downloading the same from an online document management system. At least then they can get an electronic copy.

Once the records are obtained, many vendors will then sort them into chronological order for ease in summarizing. Next, an allocator (who might be a nurse, an attorney or a claims person) creates a two to three page summary of the nature of the injury and the last two years of medical treatment. Following their record review and summarization, the allocator then develops a treatment table outlining everything the patient is likely to need for the rest of his life based on physician recommendations for specific treatments — surgeries, hospitalizations, spinal cord stimulations, etc. — as well as the routine medical care Centers for Medicare and Medicaid Services (CMS) expects to see included in the MSA.

Collecting information, sorting medical records, summarizing them, and creating a treatment table generally takes several business days, sometimes longer, depending on whether additional information is needed from the adjuster to fine-tune the MSA or if additional medical records are required. Only at this point, when the treatment and pharmacy allocations are ready for pricing, does the process begin to benefit from more automation, as most services are priced according to the workers’ comp fee schedule via an automated system or Excel worksheets. It’s quite an undertaking, but it no longer has to be.

New Era of Efficiency

New systems are being developed that use artificial intelligence to comb through medical records. Software can predict associated costs based on a wide variety of factors, such as the provider the injured worker is seeing or the costs incurred in thousands of similar cases. These tools can generate treatment plans according to hundreds of data points from within the actual claims file and corresponding bill treatment data, rather than something that a human created based on record summarization. All of this information can be pulled and analyzed in real time. It’s MSA generation at the touch of a button.

Rooting Out Errors

These efficiency improvements may sound too good to be true, but in practice, AI-based solutions are even better. One of the biggest issues, as well as being a massive cost center, is the inclusion of treatment for body parts or conditions not covered under the original claim. While it may be normal for a patient to have a change in his gait or a hip problem resulting from a knee injury, providers possibly don’t need to cover a rotator cuff in the opposite shoulder — at least not without solid documentation as to why. Yet, this is often what happens today.

Throughout the course of treatment, other ailments unrelated to the workers’ comp claim may sneak in via the billing process. These can result from a simple error by the physician’s billing service or a misunderstanding as to what’s accepted under the claim. Regardless, once that treatment is paid for somewhere along the line, it likely becomes a part of the claim for its lifetime — whether it belongs there or not.

Providers have zero desire to pay for any treatment that strays outside of the body parts initially accepted under the claim, and yet once paid, there is not much recourse. But what if a smart system, a system that had learned from thousands of other cases, was able to spot an anomaly and raise the alarm before the bill was paid and the outlier became part of the claim? This could produce significant savings annually. And, technology that can do this is being implemented in organizations now.

Moving Forward

I hopefully have shown you a cursory glimpse of some of the ways AI-based software can transform MSAs in a positive way. But I would be remiss if I made it sound like the transition to new systems is all moonlight and roses. Change is hard. People are used to doing things in very specific ways. It may take time to find and implement the right system as well as train the team on how to use it. But consider the alternative.

Organizations are hemorrhaging money and resources on MSAs, from adjusters’ time involved in the process to dollars flowing out unnecessarily in the MSA itself. AI-based solutions provide a very compelling answer, one that makes adjusters’ jobs easier and the resulting MSAs more accurate. It’s just up to you if you’re willing to take the plunge.