Catastrophic large-dollar claims moved to the front of the 2027 employer health cost trend, named a primary cause by 32% of the 112 employers surveyed by the International Foundation of Employee Benefit Plans in July and August 2026. Specialty drugs came second at 21%. The median projected increase held at 10% for a second straight year, so the mix moved while the level did not.

Key Takeaways

  • It is a share of respondents, not of trend dollars. The survey attributes no portion of the 10% to large claims, and any build-up treating the mix as weights has imported arithmetic that was never there.
  • Catastrophic rose from 31% to 32% as specialty drugs fell from 23% to 21% and chronic and mental health utilization slipped from 15% to 14%.
  • Million-dollar claim frequency is up 46% between 2022 and 2026, and claims above $2 million are up 213% since policy year 2020.
  • HHS set the 2027 ACA cost-sharing limit at $12,000 self-only and $24,000 family, which on a large loss is 0.22% of the claim.

What the 32% Actually Counts

IFEBP asked 112 corporate employers to identify the primary causes of their projected increase, and 32% named catastrophic claims. The four published drivers sum to 80%, with the remainder spread across causes the summary does not break out.

2027 leverEmployers rating it most effectiveLayer it operates inReach on a $5.45M claim
Cost-sharing initiatives19%Member, below the OOP max$12,000 maximum
Plan design initiatives15%Member and covered-service scopeExclusion only
Purchasing and provider14%Unit price, all layersFull claim, via network rate
Utilization control13%Service authorizationSite and course of care
Administration and data11%Payment integrityBilling accuracy

Nothing in the survey attributes 3.2 points of a 10% trend to large claims, and any build-up that treats the mix as weights has imported an arithmetic that was never there.

That distinction does more work than most of the coverage allows. A mention-frequency statistic reports what a benefits leader believes moved the budget. A dollar decomposition reports what actually did. The two diverge precisely when one driver is rare and enormous while another is common and moderate, which is the exact shape of the catastrophic-versus-utilization comparison.

Read as sentiment rather than as weights, the survey is still the most useful of the 2027 crop, because IFEBP samples the buyer. Segal polls carriers, PBMs and TPAs; the consultancies poll their own large-employer books. The plan sponsor is the party who saw the claim that blew up the year, and 32% of them are pointing at the same layer.

The 10% median has now printed twice, and it is structurally insensitive to the driver the same survey names. Move the largest claim in a self-funded plan from $1.5 million to $5.5 million and the median employer's projection does not move at all. A stable median across the 2026 and 2027 surveys is fully consistent with a book whose expected cost rose considerably more than 10%.

IFEBP reports what it collected, and a median is the defensible summary of 112 self-reported projections. The caution is about downstream use, because these medians travel into rate filings, budget memos and renewal negotiations as though they described a mean. Expected claims for a 500-life group are not a median. They are a mean plus a risk load, and a $5 million gene therapy claim lands squarely in the mean.

The Dollar Concentration Behind the Named Driver

The dollar-weighted evidence sits outside the release, and it argues that catastrophic claims carry far more of the 2027 increase than a 32% mention rate conveys. Sun Life's 2026 High Cost Claims report, built from more than 70,000 high-dollar claims across over 3,300 self-funded employers, put growth in million-dollar-plus claim frequency at 46% between 2022 and 2026. Blood cancers averaged $5.45 million in 2025, and the largest single leukemia claim approached $8 million.

The gradient steepens above that. Tokio Marine HCC reports claims above $2 million up 213% since policy year 2020, with 39% of stop-loss spending above $1 million attributable to children under age 10, and average severity of $1.37 million for infants under one year whose claims exceed $500,000 (Insurance Business, 2026). Jay Ritchie, who leads the carrier's accident and health group, named the mechanism: "Babies born at 21 or 22 weeks are viable now. Some of these cases involve children who remain hospitalized for over a year."

Concentration by condition points the same way. Cancer drives 35% of total paid claims in that book and cardiovascular disease another 13%, and Aon puts US million-dollar claimant growth at 45%, with catastrophic claims for rare cancers and hemophilia now landing between $4 million and $6 million. Three independent claim books, three consistent readings of the same tail.

That is what makes the mention-rate statistic and the dollar evidence point in the same direction from opposite ends. Employers are naming the driver they can see in their own experience, and the carriers holding the layer above them are reporting frequency growth in the same band. The survey understates it because a median cannot carry a tail; the stop-loss data understates the employer's exposure because it only counts what pierced the deductible.

Where the Cost Levers Stop

Set the driver list beside the strategy list and a gap opens. Employers rated cost-sharing initiatives the most effective 2027 lever at 19%, followed by plan design at 15%, purchasing and provider initiatives at 14%, utilization control at 13%, and administration or data analysis at 11%. Confidence in all three leading levers fell year over year: cost sharing from 27%, plan design from 17%, purchasing from 17%.

Cost sharing has a statutory ceiling. HHS set the 2027 ACA annual limitation at $12,000 for self-only and $24,000 for family, up roughly 13.2% from $10,600 and $21,200 (Milliman, 2026). Every non-grandfathered group plan stops collecting member cost sharing on essential health benefits at that point.

Against a $5.45 million blood cancer claim, $12,000 is 0.22% of the loss. The lever employers rate most effective is, by construction, incapable of touching the driver they rate most important, because the driver lives entirely above the point at which the lever stops working. Plan design and utilization control run into a softer version of the same wall: a gene therapy or a neonatal stay running over a year is not a utilization decision a plan can steer.

That leaves the employer's real exposure sitting in the stop-loss contract rather than the medical plan, which is where the pricing question actually lands. A self-funded sponsor buying specific stop-loss is buying protection against exactly the layer the survey identifies and the plan design cannot reach, at a price set by carriers watching 46% frequency growth in that layer. The 10% median trend describes the part of the cost the employer controls. The part it does not control is repricing at a different rate, in a market the same employer will meet at renewal, and no lever on the strategy list changes that.

Further Reading

Sources