Self-Funded Health Plan Under a 6% Cap: GLP-1 Spend, Stop-Loss Renewal, and a Union Contract

Separating the drivers

The $38.5M gross figure overstates what the plan actually bears. With three claimants above $1M and specific stop-loss at $350K, recoveries on those three alone approximate $8.5M minus three times $350K, or about $7.45M, before counting any other claimant above the attachment point. Net cost of risk is therefore far below gross claims, which means the CFO's 6% cap should be applied to net cost, not to gross claims. Setting the target on gross claims will either force unnecessary plan cuts or produce a phantom "saving."

Not controllable through plan design: the three catastrophic claimants, the timing of new high-cost diagnoses, and large-claim volatility generally. These are managed through case management, centers of excellence, and stop-loss structure, not through deductibles.

Controllable: GLP-1 utilization and eligibility criteria, rebate economics and PBM contract terms, site of service, spousal and dependent eligibility, and formulary placement. GLP-1 at $2.1M is roughly $8,000 per utilizer per year and grew roughly 250% in two years; even at only about 5% of total claims it is clearly the trend driver, and the rebate exclusion means the plan is absorbing the gross cost in a way general pharmacy spend does not.

Measurement basis

Use three-year normalized claims with a credibility adjustment rather than a single year, and report on a PMPM basis for the ongoing population plus a separately funded large-claim pool. Set the 2026 budget on net cost after stop-loss recoveries, with a stated volatility reserve. Also model a bad year: if the three catastrophic claimants run again, gross claims rise regardless of design, and the CFO needs to know that in advance.

First conclusion

Roughly a third of the 6% gap is addressable through pharmacy and vendor levers, a third through utilization management, and the remainder may simply need to be funded. Say that early.

Package

Lever	Est. annual impact	Constraint
Carve pharmacy out to a pass-through PBM with 100% rebate pass-through and a GLP-1 rebate guarantee	$700K–$1.1M	Contract transition lead time; verify the guarantee against actual GLP-1 volume, not general pharmacy
GLP-1 prior authorization with step therapy and indication-based criteria	$450K–$600K	Formulary changes generally require notice but not bargaining; check CBA drug language
Site-of-service steerage to ambulatory surgery centers with a waiver of cost-sharing differentials	$200K–$300K	Provider network disruption; member education required
Dependent eligibility audit plus working-spouse surcharge for non-union population only	$330K combined	Applying it only to non-union staff creates an equity problem; consider bargaining it in 2027
Stop-loss re-marketing at a higher specific deductible to reduce premium	$250K–$350K	Increases retained risk; requires a reserve, not a savings

Combined realistic midpoint is roughly $2.1M–$2.6M against a gap of about $2.3M on a net basis.

What I would not do

• Raise the deductible to $2,500. The CBA freezes plan design for 30% of the population through 2027, so it cannot be applied consistently; it shifts rather than reduces cost; and a $1,500 deductible is already competitive in a logistics labor market where hourly recruiting is hard. If cost-sharing must move, use coinsurance on a narrow service category, not the deductible floor.
• Reference-based pricing without wrapping it. Balance billing exposure and provider litigation risk are real, and it conflicts with the surprise-billing protections members already expect. If pursued, pair it with a member advocacy vendor and a balance-billing hold-harmless.
• Cutting GLP-1 coverage outright. Abrupt termination drives members to alternative high-cost care and creates adherence problems for diabetes patients.

Compliance and relationships

Any change to the union group requires bargaining; do not implement on the non-union group first and announce it, because that is how a mid-term bargaining demand gets filed. Any material reduction in benefits requires a summary of material modifications and plan amendment. Confirm ACA preventive coverage and out-of-pocket maximum limits before touching cost-sharing, and run a mental health parity analysis on any new prior authorization that touches behavioral health.

Reference-based pricing

The savings estimate is real but so is the tail risk. Providers have litigated against plans that pay a multiple of Medicare without adequate member protection, and members can receive balance bills that damage the employer's reputation in a tight labor market. It also interacts badly with federal surprise-billing protections if not structurally aligned. Conditions for proceeding: a written balance-billing protection for members, an advocacy vendor reachable before payment is due, a defined exception process, and a provider relations plan for the top facilities in each market. If those four cannot be funded inside the savings, do not proceed.

GLP-1 management

The trap is optimizing for the rebate rather than for net cost. The current contract already excludes GLP-1s from the rebate guarantee, which means the plan may be paying near-list for drugs that generate large rebates elsewhere in the market; the carve-out negotiation matters more than the prior authorization. Second risk: overly blunt criteria push diabetic members off therapy, and the downstream cost of uncontrolled diabetes exceeds the drug cost. Criteria should differentiate indication, not just drug class. Third: utilization is 214 people and growing; without a defined eligibility and renewal-review process, spend continues on the current trajectory.

Metrics

• PMPM and trend on a net-of-stop-loss basis, reported quarterly with a rolling four-quarter view.
• Large-claim variance against the budgeted pool; stop-loss laser exposure by named claimant.
• GLP-1: utilizers, adherence, discontinuation rate, average cost per utilizer, and gross versus net spend after rebates.
• Pharmacy: rebate yield as a percentage of eligible spend, and a specific check on the excluded categories.
• Site of service mix, avoidable emergency visits, and preventive screening completion, which is the leading indicator that cost-sharing changes are backfiring.
• Dependent eligibility and spousal surcharge uptake, with an attrition read on the hourly population.

Governance

A quarterly benefits committee that reviews claims and trend with the broker and consultant, minutes documenting the fiduciary rationale for each design decision, a formal RFP cycle for PBM and stop-loss every three years, and a reserve policy for large claims that prevents a single bad year from forcing emergency plan cuts.