CMS Administrator Dr. Mehmet Oz announced the Medicare GLP-1 Bridge on May 7, 2026: a $50 monthly copay for obesity GLP-1s, running July 1, 2026 through December 31, 2027. It is not a formulary addition. Built on Section 402 demonstration authority, the Bridge sits alongside Part D rather than inside it, and Part D sponsors carry no risk for the drugs it covers. With 2027 bids due June 1, that reshapes every GLP-1 assumption in the book.

Key Takeaways

  • $50 per month is the beneficiary copay regardless of benefit phase, and it does not count toward true out-of-pocket costs. The $245 negotiated 30-day net price does not count toward gross covered prescription drug costs either.
  • 80% NAMBA-weighted enrollment was the BALANCE model's participation threshold. CMS paused the Part D component on April 21, 2026, one day after the application deadline, and extended the Bridge instead.
  • $66.13 PMPM was the combined GLP-1 cost in Q4 2025 on Milliman's MedIntel data, up 92% year over year and the largest single drug class in Part D by total cost.
  • An 8% gap separates two defensible 2027 PMPM projections off the same 2025 data: $461 from the full-year average, $499 from the Q4 run rate. The Bridge does not resolve which basis is right.
  • $35 billion against $47.7 billion is the spread between CBO's ten-year net federal cost and the JAMA microsimulation's, on offset ratios of roughly 8% and 27.6% respectively.

What the Bridge Pays For and How

The program runs an 18-month demonstration window and covers three drug families: Wegovy from Novo Nordisk, Zepbound from Eli Lilly in the KwikPen formulation only, and Foundayo, Lilly's oral orforglipron approved April 1, 2026.

The pricing was set in the November 2025 most-favored-nation agreements. Wegovy's pre-deal list price was roughly $1,350 per month and Zepbound's exceeded $1,080. Both negotiated to $245 for a 30-day supply for Medicare and Medicaid.

Payment runs through a central processor operated by Humana on the existing Low-Income Net infrastructure. Pharmacies collect $50 from the beneficiary and are reimbursed at no lower than wholesale acquisition cost minus the copay, plus a dispensing fee. Manufacturers then remit the difference between WAC and $245 to CMS. The federal government absorbs the net cost directly.

Eligibility is narrower than a BMI threshold. It requires BMI of 30 or higher with heart failure with preserved ejection fraction, uncontrolled hypertension, or CKD stage 3a or above; or BMI of 27 or higher with pre-diabetes, prior myocardial infarction, prior stroke, or symptomatic peripheral artery disease. KFF counts roughly 14 million Medicare beneficiaries with a BMI of 25 or above on 2020 claims; the comorbidity criteria cut deeply into that.

The Risk Leaves the Bid, the Baseline Problem Does Not

For a plan actuary the Bridge is a carve-out, and the carve-out is total.

The $50 copay never counts toward TrOOP. The $245 price never enters gross covered prescription drug costs. Bridge drugs are outside the formulary, outside prior authorization, outside the risk corridors, and sponsors do not opt in. The largest single source of pharmacy trend uncertainty leaves the 2027 bid entirely.

That happened because plans refused the alternative. BALANCE would have required a $50 copay for Enhanced Alternative plans and EGWPs, $125 for others, zero catastrophic-phase cost sharing, and uniform formulary placement, with narrowed risk corridors of 2.5% rather than 5% that applied only above one standard deviation of mean utilization. Plans were already absorbing an IRA redesign in which the out-of-pocket cap drove catastrophic phase utilization roughly 22% above actuarial base cases. CMS paused the model the day after the deadline, which is what a wide miss looks like.

The carve-out does not fix the trend baseline, and that is where the residual pricing risk lives. Milliman put combined GLP-1 spend at $66.13 PMPM in Q4 2025, up 92% year over year, with non-low-income PMPM at $381 for the year and $427 in Q4 alone. Project 2027 off the annual average at a 10% trend and you get $461. Project it off the Q4 run rate and you get $499. Both bases are defensible; the 8% spread between them is larger than most GLP-1 assumption margins.

The Bridge makes that choice harder rather than easier. Reported plan-level GLP-1 PMPM should flatten for obesity indications while diabetes GLP-1 utilization stays in the benefit, so the observable series changes composition mid-window without changing definition.

The Offset Does Not Arrive Inside the Window

The economic case for the program assumes weight loss pays part of its own bill. The fiscal evidence for that happening on this timetable is thin.

Source Eligible Population Projected Users (Year 1) 10-Year Net Federal Cost Healthcare Offset Ratio
CBO (Oct 2024) 12.5 million 0.3 million (2%) ~$35 billion ~8% ($3B of $38B)
JAMA/Hwang et al. (Apr 2025) 30 million cumulative 3 million (base case) $47.7 billion net 27.6% ($18.2B of $65.9B)
AAF (Nov 2025) ~54.8M Part D enrollees 2.2 million (mid-range) ~$3.8B/year at steady state Not modeled

CBO and the JAMA microsimulation by Hwang et al. disagree by more than $12 billion on ten-year net cost because they disagree on everything upstream: 2% first-year uptake against a 10% base case, and offset ratios of roughly 8% against 27.6%. CBO projects $50 million in health savings in year one, about $50 per user, rising to roughly $650 per user by 2034. Its second decade, 2035 to 2044, is where it expects offsets to compound.

Both models assumed cost sharing well above $50. Hwang et al. used $8,412 per year net for semaglutide; CBO modeled standard Part D cost sharing. KFF survey data has about half of current GLP-1 users calling the drugs difficult to afford and a quarter very difficult. Removing that barrier for a population with higher obesity prevalence and lower disposable income than the commercial book can produce a curve steeper than either projection contemplates.

The demonstration window is the binding constraint. Eighteen months is too short to produce credible comorbidity offset data, and the clinical benefit works through the chronic disease pipeline over five to fifteen years. Duke's Sanford School summarized the near-term finding directly: the drugs worked for metabolic health but did not reduce broader healthcare use in the near term.

Then the coverage stops. When the Bridge expires on December 31, 2027, 2028 bids filed in mid-2027 must assume either a reversion to standard Part D cost sharing or a probability of extension. That is a binary regulatory outcome, and credibility standards built for statistical variation do not help price it.

Further Reading

Sources