Stop-Loss at 12.7%: The Structural Break Behind the Premium

Medical stop-loss premiums for 2026 are rising 12.7% on average across Segal's 225-plan national dataset, but the headline number understates the real problem: claimants exceeding one million dollars have grown 25% a year for four straight years (Segal, June 2026), a compounding frequency shift that has already outrun the specific attachment points self-insured sponsors negotiated in 2022 and 2023.

From reviewing stop-loss renewal packages across self-insured employers ranging from 500 to 50,000 covered lives over the past three plan years, the pattern that stands out is how quickly a single plan's own claim history stopped being predictive. A sponsor that priced comfortably at a $200,000 specific deductible in 2023, with no claims above $150,000 in its own experience, can walk into a 2026 renewal carrying one gene therapy case and several GLP-1-driven aggregate breaches its prior-period loss run gave no warning of. Segal's own year-over-year comparison makes the acceleration explicit: the 12.7% average increase for groups holding deductible levels flat is up from 9.7% in the prior cycle, and the 11.5% increase across all groups, including those that raised deductibles or moved to aggregating specifics to blunt the number, is up from 7.3% (Segal, June 2026). Carriers are not repricing a one-year spike. They are repricing a distribution that has permanently widened.

The Four-Year Run That Broke the Baseline

Segal's SHAPE claims data warehouse, which aggregates experience across the firm's national stop-loss book, shows the count of claimants exceeding $1 million growing at an average of 25% a year over the last four plan years (Segal, June 2026). Compounded rather than read as a flat annual figure, that rate means a block that produced four seven-figure claimants in 2022 should be producing roughly ten by 2026, using 1.25 raised to the fourth power, or 2.44, as the multiplier. That is a threefold escalation in the event most stop-loss contracts still treat as the rare tail case, condensed into a single renewal cycle most sponsors and brokers still budget in one-year increments.

The mechanics of that acceleration matter more than the headline growth rate. Specific stop-loss pricing at moderate attachment points has historically relied on treating claims above the deductible as a low-probability tail event: rare enough, relative to a block's total claim count, that a Poisson-style frequency assumption held reasonably steady from one rating period to the next, and pooled manual rates built on trailing three-year experience absorbed the shock cost efficiently across the whole book. A frequency process growing at a compounding 25% a year is no longer stationary within a typical pricing horizon. A rate built on trailing experience is, by construction, pricing to the average of a period during which the true underlying frequency has already moved well past that average, which is exactly the mechanism behind the gap between the 12.7% headline premium increase and the far steeper frequency curve sitting underneath it.

GLP-1s Reach the Deductible Through Accumulation, Not a Single Event

Where specific attachment points are built to catch a single catastrophic claim, GLP-1 drugs are breaching them through a different mechanism entirely: sustained, compounding annual spend. Lockton's data shows allowed per-member-per-month costs for employers covering GLP-1s for weight management rising from $2.41 to $22.59 between 2022 and 2024, an increase of roughly 840% (Lockton, via InsuranceNewsNet, 2026). Segal's 2026 Health Plan Cost Trend Survey found 60% of its client plans now offer anti-obesity medication coverage, contributing to a median 9% projected medical trend increase for 2026, the highest annual projection in more than a decade (Segal, September 2025).

Individually, injectable GLP-1s running above $10,000 per member per year rarely breach a $150,000 or $250,000 specific deductible on their own. The exposure comes from a plan holding a growing share of its population on the drugs simultaneously, a pattern John Thornton, Executive Vice President of Sales and Marketing at Amalgamated Life Insurance Company, described directly: "That sustained utilization across a growing share of the covered population compounds into aggregate pressure" (InsuranceNewsNet, 2026). That is an aggregate-layer problem, not a specific-layer one, and it explains why Segal's Q3 2026 trends report is telling sponsors to extend claims run-out periods from the standard 12/15 basis to a minimum of 12/18 or 12/24, so that delayed adjudication of accumulating high-cost pharmacy claims does not fall outside the policy period entirely (Segal, June 2026).

The pullback now underway on the employer side complicates the pricing picture rather than resolving it. Among employers with 500 or more workers, 6% dropped GLP-1 coverage for 2026 and another 5% are planning to drop it or are actively considering doing so for 2027 (CNBC, July 2026). Real-world persistence data helps explain why: adherence falls from 65% of patients still on therapy at 120 days to just 34% at one year (AJMC/Journal of Managed Care & Specialty Pharmacy, 2026). A plan that discontinues coverage or sees high attrition does not necessarily see its stop-loss exposure fall in step, because the accumulated spend from members already on therapy through much of a plan year has already flowed into the aggregate corridor calculation for that period. Coverage decisions made in 2026 will show up in aggregate stop-loss pricing with a lag, not immediately.

Gene and Cell Therapy: A Different Kind of Outlier

Gene and cell therapies break stop-loss pricing models through the opposite mechanism from GLP-1s. Where GLP-1 exposure is a volume problem spread across a growing share of a plan's population, gene therapy is a severity problem concentrated in a single patient. Per-treatment costs for currently approved therapies commonly exceed $3 million, with some running above $3.5 million (Segal, June 2026; Planned Administrators Inc.), and these are genuinely one-time costs rather than a recurring annual spend. That distinction is what makes gene therapy claims so difficult to price at the plan level: a single approval can consume an entire year's specific stop-loss budget for a mid-size employer, but the underlying disease incidence is too rare for any individual group, even one with several thousand covered lives, to generate credible claim experience of its own.

The result is that gene therapy pricing has migrated toward pooled, catastrophic-layer solutions sold alongside a sponsor's existing specific policy, rather than being absorbed into the standard specific deductible. Carriers increasingly attach laser provisions, elevated specific attachment points applied to a named individual once a high-cost condition is identified, when underwriters believe a beneficiary is likely to breach the deductible in the coming plan year. The smaller the group, the less reliable its own experience is for setting that laser, which pushes underwriting toward disease-prevalence and treatment-eligibility models rather than trailing claims history for this category specifically.

Attachment Point Adequacy for the 500-to-5,000-Life Segment

The $150,000 to $250,000 specific attachment points common in 2022 and 2023 contracts were priced against the claim distribution of that period, not the one now generating Segal's 25% annual growth figure. For a plan in the 500-to-5,000-covered-lives range, that gap has two distinct effects that compound rather than offset. First, the probability of at least one seven-figure claim breaching the deductible in a given plan year has risen in step with the 2.44x cumulative frequency multiplier shown above, which means the expected excess loss the carrier is actually paying out has grown faster than the deductible level was designed to contain. Second, claims that fall between the old attachment point and roughly $1 million, the range where GLP-1 accumulation and mid-tier specialty drug spend concentrate, now retain more variance in the aggregate layer than the corridor built around 2022-2023 experience assumed. A sponsor holding its specific deductible flat while its true underlying frequency has run to 2.44 times the 2022 level is, in effect, self-insuring a materially larger slice of expected cost than its board approved at the last renewal, even before this year's premium increase is layered on top.

How Carriers Are Repricing at Renewal

Three adjustments are showing up consistently in 2026 renewal negotiations. Laser provisions are being applied more aggressively and to a broader set of conditions than in prior cycles, moving beyond oncology and transplant history to include members with a documented history of high-cost specialty drug utilization. New exclusion language is appearing for specific high-cost condition categories, most notably around gene and cell therapy ingredient cost, pushing that exposure toward the pooled catastrophic products carriers like Planned Administrators Inc. and BCS Financial have built specifically to sit alongside a standard specific policy. And aggregate-specific blending, raising the specific deductible while narrowing the aggregate corridor, or vice versa, is being negotiated more explicitly as a single joint decision rather than two separate line items, since the frequency shift in specific claims directly changes the variance retained in the aggregate layer. Segal's report also flags a structural fee issue compounding the premium story: stop-loss interface fees currently range from $1.50 per participant per month up to $6 to $8 PPPM, and the firm is recommending sponsors push carriers toward flat monthly or quarterly fees instead of PPPM structures that scale with enrollment regardless of actual claims activity (Segal, June 2026).

What Belongs in the 2027 Baseline

The methodological error to avoid heading into 2027 budget planning is treating 2024 and 2025 experience as an anomalous peak that will mean-revert. Segal's four-year, 25%-a-year growth figure describes a trend, not a spike, and a trailing three-year average that includes 2022 and 2023 will understate the 2027 expected claim distribution by construction. A defensible 2027 baseline separates the three cost drivers by their statistical behavior rather than blending them into a single trend factor: GLP-1 exposure modeled as an aggregate-layer utilization variable tied to the plan's own enrollment percentage on the drugs, gene and cell therapy modeled as a pooled shock load using disease-prevalence and treatment-eligibility data rather than the plan's own thin claims history, and the broader seven-figure claim category modeled with an explicitly non-stationary frequency assumption on the excess layer, distinct from the severity trend applied to the body of the claim distribution. Specific and aggregate layers should be priced jointly rather than independently, since a decision to raise the specific deductible to control specific premium mechanically increases the variance retained in the aggregate corridor precisely where GLP-1 accumulation and mid-tier specialty spend now concentrate.

Why This Matters for Actuaries

The practical failure mode is a plan sponsor and its actuary treating the 12.7% headline as the full story and renewing on largely unchanged assumptions. The 25%-a-year frequency figure underneath that headline says the underlying claim distribution has already moved further than a single year's premium increase reflects, and the 2.44x compounding multiplier since 2022 means specific deductibles set at the start of that period are now covering a meaningfully smaller share of expected excess cost than they were designed to. Actuaries pricing 2027 renewals should treat 2024 and 2025 experience as the new baseline rather than an outlier to be smoothed away, separate the frequency and severity components of the trend rather than applying one blended factor, and price the specific and aggregate layers as a single joint decision. Sponsors that renew on a stale distribution will find out how exposed their attachment point really is at the worst possible moment: mid-plan-year, with a claim already in the door.