Blood cancer claim costs per employee climbed more than 70% in 2025, and neoplasms held their position as the leading driver of stop-loss reimbursements at 36% (QBE North America, via Risk & Insurance, June 2026).

That severity concentration, rather than the more widely cited jump in claim frequency, is what determines whether a carrier's reserve for the excess layer holds.

Key Takeaways

  • Blood cancers are now more than half of all neoplasm claims at the $1 million specific deductible, and their cost per employee per month rose more than 70% in a single year.
  • The highest claim QBE has recorded exceeded $7 million, running from initial oncology treatment through transplant to advanced cellular therapy.
  • Multiple myeloma alone spans $335,000 to $2.5 million in potential annual cost, a range wide enough that a diagnosis code no longer implies a narrow cost band.
  • QBE took 2025 rate increases above 20% against a market average of 9.7% that Segal reports accelerating to nearly 13% by the third quarter of 2026.
  • Predicting which member becomes a high-cost claimant is close to a coin flip, which is the constraint lasering cannot design around.

Where the Severity Concentrated

The frequency backdrop is well covered: 49% of self-funded plan sponsors now report at least one claim exceeding $1 million, up from 23% a year earlier, with frequency rising at every threshold QBE tracks from $500,000 through $2 million. One in six report a $3 million claim, overall high-cost frequency has nearly tripled in five years, and 2025 average severity ran 10% above the prior year. Fewer than 1% of covered individuals, those at $250,000 or more, account for 15% of total medical and drug spending.

"High-cost claims are no longer outliers. They are becoming a routine part of the healthcare landscape," said QBE North America President of Accident and Health Tara Krauss (Insurance Business, 2026). The excess loss factor mechanics behind the 49%-versus-23% comparison are covered elsewhere.

What moved this year is the composition. Neoplasms have topped QBE's driver list in every edition since 2023, in a band running 31% to 36%, and sit at 36% now. Inside that category, blood cancers including leukemia, lymphoma and multiple myeloma now account for more than half of all neoplasm claims at the $1 million specific deductible, with cost per employee per month up more than 70% in 2025. Myeloid leukemia showed the steepest subtype increase, and multiple myeloma cases carry potential annual costs from $335,000 to $2.5 million.

Claim category2025/2026 figureSource
Blood cancer claim cost per employee per monthUp more than 70% in 2025QBE North America, June 2026
Highest recorded single claim to dateExceeded $7 million (oncology to cellular therapy)QBE North America, June 2026
Multiple myeloma, potential annual cost$335,000 to $2.5 millionQBE North America, June 2026
Elevidys (Duchenne gene therapy), single claimantApproximately $3.4 millionQBE North America, June 2026
Nexviazyme (Pompe enzyme replacement), annual costUp to $2.1 million or moreQBE North America, June 2026
Specialty pharmaceuticals, annual cost range$90,000 (Rinvoq) to $395,000 (Opdivo)QBE North America, June 2026

The highest claim QBE has recorded exceeded $7 million, driven by oncology care progressing from initial treatment to transplant and advanced cellular therapy. That is a paid claim, not a modelled tail scenario.

Two More Tails, and What They Do to the Loss Pick

Cancer is not alone in stretching the layer. One Elevidys claimant, on the gene therapy approved for Duchenne muscular dystrophy, incurred approximately $3.4 million in paid claims, and Nexviazyme for Pompe disease can run $2.1 million or more annually for one member. Specialty pharmaceuticals range from $90,000 a year for Rinvoq to $395,000 for Opdivo, a spread sitting entirely inside what was once catastrophic territory before a plan reaches oncology at all. Total US pharmaceutical expenditures grew 12.7% in 2025 to $915.2 billion.

Preterm births form a third driver. Birth-related claim frequency more than doubled between 2024 and 2025, four of the top six pediatric claims followed extensive post-birth hospitalization, and third-party administrator payments on the costliest ranged from $10,600 to $5.6 million. Circulatory disease is the steady counterexample at roughly 12% of claims, its frequency up almost 60% since the pandemic without the same severity concentration.

The reserving consequence follows from where the specific layer sits. It is priced on the tail of the severity distribution rather than its body, so IBNR development on a handful of large claims moves a loss ratio further than a year of trend across the rest of the block. A carrier can run a well-behaved frequency-severity model across hundreds of employer groups and still see individual blocks break their loss pick.

That is because these claims are not independent draws from a stable distribution. Blood cancer, gene therapy and preterm complications cluster into the same small set of categories, and each is correlated with treatment advances that move faster than an annual severity study absorbs. A reserve anchored to a multi-year average severity trend understates the excess layer when the diagnosis mix inside that severity has shifted as sharply as a 70% single-year move in blood cancer cost, even where aggregate frequency looks stable. Pricing a specific deductible off a blended neoplasm assumption merges a subgroup whose trajectory now diverges from the rest of its own category.

The market is repricing around exactly that. QBE pushed 2025 Accident and Health rate increases above 20%, against a Segal national average of 9.7% in 2025 accelerating to nearly 13% by the third quarter of 2026 (Segal, July 2026). A carrier pricing at more than double the market's average increase while publishing the severity data the rest of the market is reacting to is catching up to its own paid claims. Capital keeps arriving anyway: the US stop-loss market was valued at $26.9 billion in 2024, and Nationwide bought Allstate's employer stop-loss segment for $1.25 billion in January 2025.

Lasering Cannot Price What It Cannot Predict

The standard response to a known high-cost claimant is lasering: carving that individual out of the specific deductible and setting a higher individual attachment point for the next year. It works as a pricing tool only if a carrier can identify in advance who will become expensive.

The research says it largely cannot. Ankit Nanda, a senior reinsurance actuary writing for the American Academy of Actuaries' journal, cited findings that the odds of correctly predicting a high-cost claimant in a policy year from past experience are no better than 50-50 (Contingencies, January 2024). His broader point is a credibility argument: "historical experience can only offer so much assistance in accurately projecting future costs," especially for the small and midsized blocks that make up most of the self-funded market, where claim volume is too thin for a reliable loss pick.

The arithmetic shows why that gap now costs more. A specific book covering an 800-life employer at a $150,000 attachment might generate on the order of $2 million to $2.5 million of annual specific premium. One claim at the $7 million ceiling, or a multiple myeloma case at the top of its $335,000-to-$2.5-million range stacked against other blood cancer claims in the same block, consumes several years of that premium in one policy period.

Against a coin-flip predictive baseline, lasering is a lagging response rather than a pricing tool. It protects the carrier from a claimant already identified, and does nothing about the next one, who is statistically just as likely to appear. What it actually does is shift realized risk back to the plan sponsor for the claimants already flagged, while leaving the carrier exposed to the undiagnosed half of next year's severity, in the categories where cost is moving fastest.

Further Reading

Sources