Pharmacy has grown into a quarter of employer health spend and is compounding at about 12% a year, so on the survey's own arithmetic the drug line supplies roughly three of the 9.2 percentage points that 127 large employers project for 2027 (Business Group on Health, August 2026). The benefit those employers cut first was GLP-1 coverage. Employer coverage for obesity fell from 72% in 2025 to 60% in 2026, a further 14% drop it next year, and not one employer in the sample is adding it.

Key Takeaways

  • 9.2% before plan changes, 8.0% after: the 120 basis points employers expect to claw back through benefit design is narrower than the roughly 150 points they budgeted for 2026, even as the gross projection rose 70 points.
  • 72% to 60%: employer coverage of GLP-1s for obesity fell 12 points in a single year, with a further 14% dropping it for 2027 and none in the survey adding it.
  • 60.9% one-year persistence among weight-loss GLP-1 initiators in the first half of 2024, up from 33.2% in 2021, which retires the drop-off assumption sitting underneath most 2023 and 2024 budget models (JMCP, 2026).
  • 70% named cancer a top cost driver this year, up from 58%, and oncology is not a line an employer can exclude by fiat the way a therapeutic class is.

What the 9.2% Measures, and What the 8% Assumes

The headline is a before-and-after pair. Employers put the 2027 median increase at 9.2% before plan changes and about 8% after the changes they have already settled on. The equivalent 2026 pair was 8.5% and 7%. Read across the two years, the gross projection rose 70 basis points while the relief employers expect from benefit design narrowed from roughly 150 basis points to roughly 120.

That movement is the more informative one, and it is invisible in coverage that leads on 9.2%. Plan design is a stock, not a flow. Every year a sponsor raises a deductible, narrows a formulary or tightens an eligibility rule, the next year's version of that lever has less travel left. A 120 basis point haircut against a 9.2% gross number is a plan sponsor working harder for less.

The survey's own forecast history argues even the gross number is generous. Business Group on Health reports actual costs exceeding employer projections in each of the past three years, with the miss widening. Ellen Kelsay, the group's president and chief executive, put it plainly: "This pattern suggests that current forecasts for 2026 and 2027 may actually be too optimistic" (Healthcare Dive, August 2026).

The spread against the other 2027 surveys is mostly definitional rather than substantive.

Survey 2027 before plan changes 2027 after plan changes Respondents Covered population
WTW (August 2026) 11.1% 9.7% 471 employers, 100 to 25,000+ workers 7 million employees
Aon (August 20, 2026) 9.5% Not published 1,100+ employers 7.9 million employees, $135B 2026 spend
Business Group on Health (August 25, 2026) 9.2% 8.0% 127 employers 11 million people globally

WTW printed 11.1% before plan changes against Business Group on Health's 9.2%, but on a comparable after-redesign basis the two are 9.7% and 8.0%, a gap explained largely by respondent mix and employer size rather than by disagreement about trend. For an actuary setting a 2027 self-funded budget rate, 8% is a conditional forecast: conditioned on levers with progressively less room, and drawn from a series biased low three years running. That is the same credibility question the site worked through on the manual side of PwC's 9% group medical benchmark.

The Coverage Drop and the Pool It Leaves Behind

Employers keeping the benefit are tightening entry rather than exiting, adding biometric validation and stricter eligibility criteria (Reuters, August 25, 2026). Both moves reduce the count of covered members. Neither reduces cost per covered member, and the second raises it.

The GLP-1 budgets that came in under projection in 2023 and 2024 did so largely because members quit. That is no longer the behavior. One-year persistence among initiators of high-potency, weight-loss-indicated GLP-1 receptor agonists rose from 33.2% in 2021 to 60.9% in the first half of 2024, with tirzepatide initiators at 64.8% (Journal of Managed Care & Specialty Pharmacy, 2026). Holding new starts constant, a cohort persisting at 61% rather than 33% carries roughly 1.8 times the annualized drug cost.

That ratio is the pricing problem. A prior-year GLP-1 per-member-per-month figure pulled off 2023 or 2024 experience embeds the old drop-off rate, and trending it forward at a pharmacy trend factor compounds a base that was low for a structural reason no longer holding. The correction belongs in the persistence assumption inside the utilization model, not in the trend factor applied on top of it. Applying a 12% pharmacy trend to a stale base produces a number that looks defensible on the worksheet and understates the class. The site tracked the same gap between assumed and demonstrated savings in GLP-1 loads in 2027 bid development.

Unit cost makes it unforgiving. Wegovy lists at $1,349.02 per month and Zepbound at $499. At list, a member persisting a full year on Wegovy runs about $16,190 before rebates, within a few hundred dollars of the $17,562 total plan cost Aon measured per employee across all benefits in 2026. One persistent claimant is close to a whole average employee.

Screening on biometrics does not remove a random slice of that population. It removes the marginal user: lower BMI, fewer comorbidities and, on the persistence evidence, the shortest expected therapy duration. What survives the screen is the higher-acuity member most likely to stay on drug for years. Utilizer counts fall while per-utilizer cost of the retained cohort rises, so a plan modelling the change as a simple reduction in covered lives books a saving it does not collect.

The Cost Line the Coverage Decision Does Not Touch

Cancer was named a top cost driver by 70% of respondents this year, up from 58%, with musculoskeletal and cardiovascular conditions behind it (MedCity News, August 2026). Oncology, high-cost claimants and specialty infusions are not benefit-design problems in the way a weight-loss drug is. An employer can exclude a therapeutic class by fiat. It cannot exclude a stage-three diagnosis.

So the 14% dropping GLP-1 coverage are cutting the largest line they can actually steer, against a trend whose fastest-growing components sit outside their control. That is defensible inside a single budget year and weak across a projection horizon, because the saving is a one-time level shift while the drivers behind it compound. The catastrophic claims driver the buyer-side surveys keep flagging is the part of the 9.2% that no formulary decision reaches.

It is also not a saving that stays where it was booked. Members do not leave the plan when the coverage does. Weight regained after therapy stops re-presents as cardiometabolic and musculoskeletal claims, on the medical side of the ledger, in a later experience period. A plan removing the benefit for 2027 books the pharmacy reduction immediately and carries an unquantified medical offset at an unknown lag. Neither the size nor the timing is in the survey, and neither is in most 2027 budget rates.

For the 60% still covering the therapy, the complication runs the other way. No employer in the sample is adding coverage, so members who want the drug face a shrinking set of plans to be in. Across a hiring cycle that sorts them toward the plans that cover it, and the retained pool is one whose utilizers are self-selected on top of an eligibility screen already selecting for persistence. A pharmacy trend assumption for those plans has to carry both effects at once, and a manual trend drawn from a mixed book of covering and non-covering employers carries neither.

Further Reading

Sources