“Holy moly, 100 grand. This is as much the claims handling process as anything.”
- Senior Manager, Risk & Insurance, North America, at a global services contractor with more than 50,000 employees
An industry POV for Risk Managers, Directors of Risk Management, and Workers’ Compensation Program leaders at self-insured employers.
By Nathan Gunn, MD, CEO and Co-Founder, SecondLook Health
June 15, 2026
A self-insured employer recently received an MSA allocation projecting $147,310 in lifetime future medical costs. A clinical re-read brought the same allocation in at $47,310.
How is a $100,000 swing on a single MSA possible, where did the over-allocation come from, and what role does AI play in closing it?
The answer is that MSA accuracy depends on a causation analysis the typical cost-projection workflow doesn't perform. Doing that causation analysis on every MSA isn't practical the way MSAs are prepared today. SecondLook Health's clinical AI is what makes it practical.
The employer is a Fortune 500 services contractor with a self-insured high-deductible workers’ compensation program. Late in 2025 the employer’s risk leadership engaged SecondLook Health to re-read an MSA allocation prepared by their existing MSA vendor through their TPA. The original allocation came in at a projected $147,310.
The claimant was a 62-year-old male injured in a severe rear-end motor vehicle collision while on duty. Post-MVA imaging showed no acute spinal injuries: CT and X-ray returned no acute changes, and the only documented injuries were a closed head injury and a right elbow sprain. Three months later, an MRI documented multilevel degenerative spine disease consistent with chronic inflammatory endplate disease, and a lifelong anomaly that accelerates degeneration. The treating spine specialist eventually recommended a lumbar interbody fusion with posterior screw fixation. The original MSA included that fusion, plus post-operative rehabilitation, as a future MVA-related medical cost. That recommendation is where the original allocation stopped asking questions, and where the re-read started.
Leta Sharkey-Laugle, MS, NCC, CRP, CLCP, MSCC -- Senior Medicare Settlement Consultant at SecondLook Health -- reviewed the same medical record using SecondLook’s clinical intelligence platform. Her revised allocation came in at $47,310.
“Many MSA vendors would add in a spinal fusion and subsequent rehab because the doctor said so. But if you read the doctor’s reports, he doesn’t relate it causally. Even his diagnosis doesn’t attribute it to the injury. His diagnosis says spondylosis. The literature is real clear: post-traumatic fusion requires a spinal fracture, neural compression after fracture, or progressive significant acute neurological impairment. None of those apply here.”
- Leta Sharkey-Laugle, MS, NCC, CRP, CLCP, MSCC, Medical Settlement Advantage Expert Inc.
Leta’s read of the record is what a credentialed reviewer does when the workflow asks the causation question. The reason this doesn’t happen on most MSAs is structural.
"On my side of the world, it's a little bit like smoke and mirrors. I want a review to make sure [the assessment is] legitimate. I just literally say that, but I don't know the quality of what and who's doing that. I'm relying on that third party administrator."
- Senior Manager, Risk & Insurance, North America, at a global services contractor with more than 50,000 employees
MSA preparation today is dominated by cost projection. The vendor pulls the medical record, identifies the documented treatment plan, and prices it forward using fee schedules, life expectancy tables, and pharmacy projections. The methodology is well established and the math is accurate. What sits upstream of this math is more important.
When the medical record contains a documented future treatment recommendation, the MSA vendor has to decide whether to include it. The default is to include what the treating physician recommended. From the vendor's perspective, the downside risk for under-allocating and getting flagged by CMS is larger than the cost of over-allocating, even when the over-allocation runs into six figures and the employer pays.
Even when an employer recognizes the gap, the standard alternative is to retain a credentialed clinician to read the chart manually. A complex multi-provider record review cost tens of thousands of dollars per case. That cost is hard to justify on every MSA, so the work happens on the ones that get flagged and not on the ones that don’t. Six-figure over-allocations on unflagged cases keep getting paid.
Three things make the SecondLook platform different from a generic chart reader or a literature search wrapper, and all three were active on this case.
On this case, SecondLook’s AI did not decide whether the fusion was compensable, did not include or exclude items from the MSA, did not communicate with CMS, and did not sign anything. The benefits determination came from credentialed professionals.
A $100,000 swing on one MSA is meaningful for a self-insured employer. The bigger number is what happens when the same pattern shows up across a book of business.
Medical costs account for 60-90% of total losses on the high-severity workers' compensation claims that drive MSA work, with the share rising as claim size grows.¹ These are exactly the claims that fall under Medicare's WCMSA review threshold. The Centers for Medicare and Medicaid Services issued 14,862 WCMSA determinations in fiscal year 2024, with an average proposed allocation of approximately $70,887.²
Published biomechanical literature establishes that disc injury from motor vehicle collisions is rare, and that disc herniations develop as fatigue injuries from cyclic loading over years rather than acute trauma.³ The North American Spine Society guideline on degenerative lumbar spondylolisthesis and the AHRQ systematic review both explicitly exclude traumatic from degenerative classifications.⁴⁻⁵
Every MSA in your book where a documented treatment recommendation overlaps with a degenerative finding sits at this intersection. A self-insured employer with five to ten MSAs running through their program in a year is exposed to a number that adds up fast.
“Oh, my God. I never realized I had influence on this. Before, I was always like, well, they just do it. I have a vendor, they know what they’re doing.”
- Senior Manager, Risk & Insurance, North America, at a global services contractor with more than 50,000 employees
The dollar gap between an MSA that engages the causation question and one that defaults to including what the doctor recommended is large enough to show up on a single case and reproducible enough to show up across a book. The structural fix is to put a credentialed clinical reviewer, supported by a physician-led clinical intelligence platform that makes the work feasible at scale, into the workflow before the math gets done.
Three concrete actions a Risk Manager can take this quarter, in order of effort.
1. Pull one open MSA from your book that matches the four signals above and book a 15-minute strategy conversation. SecondLook will identify documented future treatment recommendations, examine the causation chain in the medical record, and return the current peer-reviewed literature on any contested allocation.
2. Audit MSAs in your book prepared more than twelve months ago for documented surgical recommendations in claimants with pre-existing degenerative findings. This is the highest-yield subset for a re-read.
3. If you advise self-insured employer clients as a risk consultant or benefits consultant, request a client-facing checklist for evaluating clinical AI in workers’ compensation by emailing support@secondlookhealth.ai.
Credentialed MSA professionals (MSCC, CMSP, CLCP) who want to use SecondLook directly for their own signed work can start a free case.
1. National Council on Compensation Insurance. 2025 State of the Line Guide. Medical share of losses by claim severity threshold, Accident Years 2004 to 2024. https://www.ncci.com/SecureDocuments/SOLGuide_2025.html
2. Centers for Medicare and Medicaid Services. Workers’ Compensation Medicare Set-Aside (WCMSA) Fiscal Year Statistics: 2020 to 2024. Published November 2024. FY 2024 figures: 14,862 determinations, average proposed WCMSA $70,887. https://www.cms.gov/medicare/coordination-benefits-recovery/workers-comp-set-aside-arrangements
3. Kent R, Cormier J, McMurry TL, Ivarsson BJ, Funk J, Hartka T, Sochor M. Spinal injury rates and specific causation in motor vehicle collisions. Accident Analysis and Prevention. 2023 Jun;186:107047. doi:10.1016/j.aap.2023.107047. PMID: 37003164. https://pubmed.ncbi.nlm.nih.gov/37003164/
4. North American Spine Society. Clinical Guidelines for Diagnosis and Treatment of Degenerative Lumbar Spondylolisthesis. Definition excludes traumatic spondylolisthesis. https://www.spine.org
5. Agency for Healthcare Research and Quality (AHRQ). Systematic review distinguishing degenerative from traumatic spine disease classifications. https://www.ahrq.gov
6. U.S. Department of Veterans Affairs, Office of Community Care. Clinical Determinations and Indications: Spinal Fusion Surgery (IVC-CDI-00054). Criteria distinguishing degenerative from post-traumatic indications for lumbar fusion. https://www.va.gov/COMMUNITYCARE/docs/providers/CDI/IVC-CDI-00054.pdf
The Untapped AI Opportunity in Workers’ Compensation
