California spent hundreds of millions on weight-loss drugs. It has not shown the public what it bought.

I. What the state actually spent, and on whom

California’s Department of Health Care Access and Information — the state’s health-payments data authority — has published a data brief tracking GLP-1 prescriptions for weight loss from 2018 through 2023. To isolate likely weight-loss use, HCAI built a cohort that excluded patients coded with diabetes, then tracked prescriptions and cost for five GLP-1 products within that population across both Medi-Cal and commercial insurance.

PROVED Within that constructed cohort, Medi-Cal gross pharmacy cost for GLP-1 prescriptions reached $416.8 million in 2023. Commercial-payer cost in the same cohort reached $405.2 million. The combined 2023 figure across both payer types was approximately $822 million, gross and pre-rebate. Prescription volume in the cohort grew from roughly 20,000 in 2018 to more than 700,000 in 2023 — a level HCAI itself describes as over thirty-four times the 2018 figure.

Three qualifications belong beside those numbers, and a serious accounting cannot omit them.

First, this is not a billion-dollar figure, and it is not a diagnosis-free figure. HCAI removed diabetes-coded claims to approximate weight-loss use. It did not, and could not, screen for hypertension, prediabetes, obstructive sleep apnoea, cardiovascular risk, or the qualifying BMI thresholds under which Wegovy, Zepbound and Saxenda carry FDA approval for chronic weight management. Obesity is itself a recognised chronic disease, a point California’s own legislative findings expressly recite. NOT ESTABLISHED Nothing in this record supports the claim that these prescriptions were cosmetic, or that the people who received them had no medical reason to. The correct claim is narrower and, properly understood, more consequential: California spent this money on an FDA-approved indication, through a benefit it had voluntarily elected to offer, and cannot show what the expenditure achieved.

Second, the $822 million is gross, not net. HCAI’s documentation states plainly that its cost figures do not reflect manufacturer rebates, coupons or other discounts. DOCUMENTED The Legislative Analyst’s Office has been reported to find federal rebates running roughly 40 to 50 per cent of gross Medi-Cal pharmacy spending in recent years, with state supplemental rebates adding a few further points; this review did not confirm that range against the brief itself. What the LAO does establish is that drug-level rebate data is confidential by federal design and that net spending trends at drug level are uncertain. Net state cost is very likely a fraction of $822 million. The state has not published the fraction.

Third, resist collapsing multiple fiscal years into an undifferentiated “billions” figure absent a published cumulative accounting. INFERRED A multi-year cumulative Medi-Cal expenditure across the high-growth years plausibly runs to several hundred million dollars net. This review found no single published state document totalling it, and declines to manufacture a round number the state itself has not supplied. A critique containing a figure the state can show is unsupported stops being a critique and becomes the story; the accounting gap then disappears from view. The gap is the finding, and it survives scrutiny only if every number around it does.

A methodological note on 2023 as an endpoint

HCAI’s diagnosis-filtered analysis covers only 2018 through 2023. That is a limit on HCAI’s published, diagnosis-linked product — not a limit on the existence of California drug-claims data after 2023, and the distinction matters enough to state carefully, because conflating the two would overstate the case.

California’s Healthcare Payments Data system, the all-payer claims database the brief was drawn from, continued collecting claims past 2023, with non-public research files available to qualified applicants through a formal application process. DOCUMENTED Separately, the federal Centers for Medicare & Medicaid Services publishes State Drug Utilization Data — prescription counts and gross reimbursement by state, quarter and national drug code — with public releases reported to cover California through 2024 and into 2025.

That federal data cannot isolate weight-loss use the way HCAI’s method can, because it carries no beneficiary diagnosis: a California record for Ozempic or Mounjaro cannot be assumed to represent obesity treatment rather than diabetes treatment. But for Wegovy, Zepbound and Saxenda — brands whose primary indication through most of this period was chronic weight management — it can establish California-specific prescription volume and gross Medicaid reimbursement for 2024 and 2025. This review has not performed that extraction, and says so rather than implying a post-2023 series it does not have.

INFERRED One complication cuts against brand identity as a proxy for indication: both Wegovy and Zepbound acquired non-weight-loss FDA indications during exactly this window — cardiovascular risk reduction in March 2024, obstructive sleep apnoea in December 2024. A 2025 Wegovy claim is therefore a weaker signal of weight-management use than a 2023 one. That is an argument for diagnosis linkage, not against using later data.

II. Why this was not a marginal line item

PROVED The LAO’s April 2025 brief — the most rigorous public accounting of Medi-Cal drug spending available — found that Medi-Cal pharmacy spending overall nearly doubled between 2018–19 and 2023–24, and attributed nearly half of that growth to a cluster of drugs treating diabetes, obesity and inflammatory disease.

DOCUMENTED Within that cluster, the LAO is reported to have isolated GLP-1 agonists as the most notable increase, alone comprising roughly a quarter of total pharmacy spending growth over the period; and to have projected gross Medi-Cal pharmacy spending at $19.4 billion for 2024–25 rising to $20.6 billion for 2025–26, attributing part of the increase in average drug cost specifically to specialty diabetes and obesity medications. This review verified the cluster finding directly and records the GLP-1-only share and the dollar projections as reported rather than confirmed.

DOCUMENTED The LAO also situated California’s choice comparatively: only twelve other state Medicaid programmes covered GLP-1s for obesity at the time of its analysis; Medicare historically excluded them; most private plan enrollees lacked comprehensive coverage. California’s Medi-Cal population is reported to have an adult obesity rate near 40 per cent, higher than Californians with private insurance or Medicare.

INFERRED The state made an unusually generous and unusually exposed policy choice, covering a population with unusually high clinical need. That combination made California more vulnerable to the cost trajectory that followed — and it is precisely the combination that makes an outcomes accounting more valuable, not less. It also created something rarer: one of the largest publicly financed natural experiments in anti-obesity pharmacotherapy in the United States, in a low-income population that clinical trials systematically under-represent.

California’s public record can tell us with increasing precision what it bought. It is far harder to find a comparable public record showing what health improvement it purchased.

III. The termination: precise, deliberate, and cost-driven

PROVED The elimination of Medi-Cal’s GLP-1 weight-loss benefit was not a drift or a lapse. It was a specific, budget-enacted decision executed through a documented regulatory process, and the department’s own pharmacy bulletins record it in operational detail.

The May 2025 budget revision first proposed elimination, projecting $85 million in General Fund savings for 2025–26 growing to $680 million by 2028–29. The enacted 2025–26 Budget Act revised that terminal projection upward to $790 million by 2028–29 and ongoing — a revision worth noting on its own, since it reflects the administration’s own escalating estimate of what continued coverage would have cost. Anyone citing “the savings figure” should say which instrument it came from; the two differ by $110 million a year.

PROVED Three products carried a Contract Drugs List entry with a Code I diagnosis restriction of Chronic Weight Management: Wegovy, Zepbound and Saxenda. Effective 1 January 2026 all three were removed from the list entirely. The January 2026 monthly bulletin records the removals by generic name — semaglutide, tirzepatide, liraglutide, each “Removed from the CDL.”

The enforcement mechanism is a rejection code. Claims for those three now deny with Reject Code 70 — Product/Service Not Covered, regardless of indication, and previously approved prior authorisations were deemed expired rather than allowed to run to their end dates. Two narrow case-by-case exceptions survive: Wegovy for noncirrhotic metabolic dysfunction-associated steatohepatitis (MASH) or cardiovascular disease, and Zepbound for obstructive sleep apnoea. Members under 21 may have a weight-loss request reviewed for medical necessity under federal EPSDT requirements. Seven other products — Ozempic, Rybelsus, Mounjaro, Victoza, Byetta, Bydureon and Trulicity — remain listed under a type 2 diabetes restriction, denying with Reject Code 80 when that restriction is not met.

DOCUMENTED DHCS is reported to have executed the change through State Plan Amendment 25-0029, submitted to CMS on 30 December 2025 under Section 1927 of the Social Security Act — the provision classifying weight-loss agents as an optional Medicaid benefit a state may decline to cover — with member notification letters issued sixty days ahead. This review did not locate the amendment in a state or federal filing repository. The operational facts above do not depend on it.

NOT ESTABLISHED The stated reasoning was uniform across every bulletin examined: pursuant to the enacted 2025–26 State Budget, to reduce pharmacy spending, improve programme integrity, and ensure continued, equitable access. Nothing in the public record suggests the termination followed a finding that the drugs did not work. The record shows a fiscal retreat from an optional benefit whose cost trajectory had outrun the budget built to absorb it, executed competently and on notice.

IV. What California has not published

This is the centre of the story, and it should not be mistaken for a claim that no clinician monitored a patient or that GLP-1 drugs lack proven efficacy. Trial evidence for semaglutide and tirzepatide’s effect on body weight is strong and well established. California’s own legislative findings recite obesity’s links to diabetes and cardiovascular disease. None of that is in dispute.

What this review could not locate — after examining HCAI’s public data products, the LAO’s dedicated pharmacy brief, DHCS’s member-facing Medi-Cal Rx bulletins, and the public record surrounding the elimination — is a Medi-Cal-specific published evaluation connecting the terminated benefit to measured patient outcomes.

Specifically, this review found no state-published figures — for the Medi-Cal population that received this benefit — on any of the following:

• average or median weight change achieved;
• the proportion of beneficiaries reaching 5, 10 or 15 per cent weight reduction;
• treatment persistence at six, twelve or twenty-four months;
• rates of discontinuation, or weight regain following discontinuation;
• progression from prediabetes to diagnosed diabetes among treated beneficiaries;
• cardiovascular events among treated beneficiaries;
• changes in antihypertensive, antidiabetic or lipid-lowering medication use following initiation;
• any downstream Medi-Cal medical-cost offset attributable to the benefit; or
• net cost per beneficiary after rebates.

HCAI’s own description of its data source is instructive for what it concedes rather than what it claims: the underlying dataset is administrative claims and encounters generated through transactions among payers and providers. That is a record of what was billed, not a record of what happened to the patient. It is an excellent instrument for measuring utilization and gross cost. It was never designed to answer a clinical-outcomes question, and HCAI does not claim it does.

INFERRED The finding, stated at its proper weight. The gap is not that California stopped collecting drug-claims data, and not that no post-2023 pharmacy figures exist — both state and federal claims infrastructure continued operating past 2023. The gap is narrower and more consequential. California has published extensive, granular utilization and cost data: what was dispensed, to how many people, at what price. It has not published a comparable clinical accounting of what those dispensations achieved.

The absence of such a document is not conclusive proof that no internal analysis exists within DHCS. It is proof that none has been made available to the public, to the Legislature’s own nonpartisan analyst, or — on the evidence available — even contemplated by the LAO’s own transparency recommendations.

That last point deserves more weight than it has received. The LAO asked DHCS to report more about utilization, cost and rebates. It did not ask for clinical outcomes at all. That is a sound recommendation on its own terms — those were the gaps blocking the analyst’s budget work. But it means the accountability framework the Legislature’s own analyst proposed going forward leaves the effectiveness question entirely outside it. Had the recommendation been adopted in full, California would have obtained better numbers about money and still none about health.

The state’s most complete public analysis of this programme was finalised five months after the programme was terminated — and it answered questions about volume and price.

What could have been answered from data already held

In fairness to the state, an honest account must concede what a claims-based evaluation could not have delivered. Pharmacy and medical claims contain no weights. A rigorous weight-outcome analysis needs measurements from clinical records, which live in managed-care plans and clinic systems rather than in the payment layer. Building that linkage involves data-use agreements, privacy safeguards and plan cooperation. Attribution is hard too: members enter and leave Medi-Cal, prescriptions are abandoned, weight changes for reasons unrelated to any drug.

But that concession has a boundary, and the boundary is the point. INFERRED Three of the questions listed above need no clinical record at all. Treatment persistence is a refill-pattern calculation. Progression to a diabetes diagnosis is a diagnosis-code question. Downstream utilization is a claims-linkage question. Those are ordinary analyses of data California already holds — and the state performed a considerably more elaborate exercise, constructing a diabetes-exclusion cohort across five products and six years and mapping it by county, to produce the cost figures it did publish.

The capability was demonstrated. It was directed at what the benefit cost rather than at what it did.

The people who were mid-treatment on 31 December 2025

Every account of this decision so far has been fiscal. There is a clinical account too, and it has had almost no attention.

GLP-1 receptor agonists do not work like a course of antibiotics that is finished and put down. They suppress appetite and slow gastric emptying for as long as they are taken, and the physiological drivers of weight regain — lowered resting energy expenditure, altered satiety signalling, restored hunger — return when they stop. The clinical literature on discontinuation is consistent on the direction of travel: a substantial share of lost weight comes back, over months rather than years.

So on 31 December 2025, when prior authorisations were deemed expired, a population of Medi-Cal members who had been treated — some for years, some of whom had reached clinically meaningful reduction — reached the end of coverage. Some will have had a clinician who could document a covered indication. Some will have paid cash. Some will have stopped. How many fell into each group is not published either.

INFERRED That matters fiscally as well as clinically, because the savings projection assumes the spending simply ceases. If a share of those members regain weight and progress to diabetes, the state pays for that too — through the GLP-1 coverage federal law requires for diabetes, and through the rest of the disease. Whether the projected $790 million is a net saving or a deferred cost is precisely the question an outcomes evaluation would have informed.

There is one further consequence worth naming because this site has documented it. When coverage for an expensive medicine ends and demand does not, the demand goes somewhere. FDA is explicit that compounded drugs are not FDA approved, which means the agency does not review their safety, effectiveness or quality before marketing — and FDA has accumulated over a thousand adverse-event reports involving compounded semaglutide and tirzepatide. A benefit termination is not only a budget line; it is a redirection of patients toward a market with materially different safety assurances. Our patient guide on compounded semaglutide and tirzepatide exists because of this dynamic.

Why the answer matters in both directions

This is not a rhetorical question with one acceptable answer, and it would be dishonest to frame it as one.

Both readings are consistent with everything California has published. That is the problem. A programme of this size should not be able to end without the public learning which description is true.

V. The sequence that makes this consequential now

The story would be an ordinary budget retrenchment if it ended on 1 January 2026. It does not.

DOCUMENTED The bill’s scope is reported to have narrowed substantially between introduction and passage. As introduced it would have required CalPERS — covering roughly 1.3 million public employees, retirees and dependents — to cover at least one GLP-1 for weight management. The CalPERS board is reported to have formally opposed that provision in June 2026, citing a projected first-year statewide premium increase of $437.3 million, about $28 per member per month, with a General Fund share near $187 million and no cost-containment mechanism. That mandate is reported not to have survived amendment.

DOCUMENTED The version that advanced instead directs the state to seek a partnership with a manufacturer to increase competition and lower GLP-1 prices through CalRx — the programme that previously brought a state-branded insulin to market — with a directive to negotiate at or below what Medi-Cal paid in 2025. Reporting notes the insulin effort took roughly three years, and that a GLP-1 equivalent would likely be a multi-year endeavour.

INFERRED The sequence bears restating plainly, because its logic is the actual news. California financed broad access. Utilization and gross cost rose faster than the budget could absorb. California terminated the benefit on explicit fiscal grounds. And within the same legislative session, California began building the machinery to re-enter the same drug class at a lower price.

California financed broad access, terminated the benefit on fiscal grounds, and in the same session began building the machinery to re-enter the same drug class at a lower price.

INFERRED If the CalRx route is the mechanism, timing matters to the argument. An insulin partnership on a roughly three-year arc means the members who lost coverage on 1 January 2026 are not the members a CalRx product would serve. The gap between termination and replacement is a period in which the affected population has neither the benefit nor its cheaper successor — and nobody is currently measuring it.

That is not necessarily incoherent policy. A state may reasonably conclude that price, not clinical value, was the binding constraint. But a Legislature and a public asked to evaluate the next iteration of GLP-1 policy are entitled to know what the first iteration accomplished. On the present record, they do not.

VI. What this is not

This review is not an argument that obesity is a lifestyle condition undeserving of pharmacological treatment, that GLP-1 medications are experimental, or that Medi-Cal beneficiaries who received them did so improperly. NOT ESTABLISHED None of the primary sources examined here — not HCAI, not the LAO, not DHCS — makes any such claim, and this review adopts none of them.

The claim is narrower, and survives scrutiny precisely because it is narrower. A government that spends hundreds of millions of dollars on a discretionary clinical benefit, tracks that spending in granular administrative detail, and then withdraws the benefit for cost reasons owes an accounting — to the Legislature that appropriates the money and the public that bears it — of what the expenditure purchased in health. California has met the first obligation with unusual rigour. On the public record examined here, it has not yet met the second.

California’s GLP-1 benefit, by the record

Swipe the table sideways for the remaining columns

All cost figures are gross, pre-rebate. HCAI states its figures do not reflect rebates, coupons or discounts. Drug-level rebate data is confidential by federal design, so net state cost for GLP-1s specifically is not publicly quantified. HCAI’s cohort excludes patients coded with diabetes to raise the likelihood that remaining prescriptions reflect weight-loss use; it does not establish that those patients had no other qualifying condition. The final row is a finding about the public record, not an error state and not a claim that the benefit failed.

Method, sources, and what remains open

Figures were taken from the named California publications rather than from secondary summaries, and each is identified below with the document it came from.

Swipe the table sideways for the remaining columns

Reported, not verified here

Each item below was described to this review or appears in secondary reporting and could not be confirmed against a primary document for this page. They are listed so a reader can see what was excluded and why, rather than finding them stated as fact: State Plan Amendment 25-0029 and its 30 December 2025 submission date; the CalPERS board’s $437.3 million premium projection and the amendment history of SB 1089; the count of other state Medicaid programmes covering these drugs for obesity; the Medi-Cal adult obesity rate; the LAO’s Wegovy-specific FY 2023–24 estimate; the 40–50 per cent rebate range; the attribution of roughly a quarter of pharmacy-spending growth to GLP-1s alone; the $19.4 billion and $20.6 billion gross pharmacy projections; the CalRx insulin timeline; the Medi-Cal Rx fee-for-service consolidation date; the two FDA indication expansions in 2024; the Senate vote of 39–0; and the CHBRP independent bill analysis.

Figures deliberately excluded

• An average annual cost per member figure circulated in draft materials could not be verified against HCAI’s published output. Excluded.
• A DHCS GLP-1 savings estimate attributed to a November 2025 Medi-Cal document could not be verified; a similar figure appears in May 2025 materials attached to an unrelated line item. Likely misattribution. Excluded.
• A unique-beneficiary count appearing in secondary reporting is not interchangeable with the prescription count and was not verified against the cohort definition. Excluded, and it must not migrate into a caption.
• “California spent billions on cosmetic weight loss.” Unsupported on both counts. Excluded.

What would close this

Four things, in ascending order of difficulty, three of which require no new data collection:

1. Extend the published series to 2024 and 2025. The claims exist and HCAI has already built the cohort definition. This alone would establish whether a cumulative figure in the billions is real.
2. Publish treatment persistence from the same claims. Refill patterns answer how long members actually stayed on therapy without touching a clinical record.
3. Link pharmacy to medical claims for progression to diabetes, cardiovascular events and downstream utilization — the analysis The Obesity Society said was missing.
4. Report net cost after rebates, which is what the LAO asked the Legislature to require annually.

A fuller reconstruction would also require direct extraction of California rows from the federal State Drug Utilization Data files for 2024 and 2025, isolating the three weight-management brands by national drug code; a formal application to HCAI’s data access programme for a diagnosis-linked file extending the methodology into 2024; and a public records request to DHCS for any internal analysis that exists but has not been published.

None of that changes the central finding. Reconstructing later utilization would sharpen the spending side of the ledger. It would not supply the missing half — because reimbursement data, however current, cannot tell us whether a beneficiary who received Wegovy in 2024 lost weight, kept it off, or avoided a diabetes diagnosis as a result. That is the accounting California has not published, for any year of the programme, and no volume of additional utilization data closes it.

Written by Kanwar Partap Singh Gill, MD, family medicine physician in Fresno, California, USA · Published 5 September 2026 · Evidence current through 5 September 2026