Medicare GLP-1 Prescriptions by State: 2024 Part D Claims Data

By The RX Index Editorial Team · Data year: 2024 · Dataset version: 1.1.0 · Assembled: July 31, 2026 · Last verified: August 1, 2026

Primary source: CMS Medicare Part D Prescribers — by Geography and Drug, 2019–2024 annual files. This page contains no advertising, affiliate links, sponsored placements, provider recommendations, or lead-routing elements.

The Medicare GLP-1 prescriptions by state data show 21,825,833 Medicare Part D claims in 2024 for the included standalone GLP-1 receptor agonists and tirzepatide, carrying $27.503 billion in gross drug cost. California prescribers accounted for the most — 1,865,079 claims — followed by Texas, New York, and Florida.

The second comparison changes the picture. California was first by raw claims but 31st when the numerator was divided by all published Part D claims attributed to prescribers in the state. Alaska was 50th by raw claims and first by that within-file share. The raw totals spanned 68.8-fold; the within-file shares spanned 1.71-fold. That inversion is the single most important thing on this page.

Across all 50 states and the District of Columbia, that within-file spread narrowed in every annual file from 2019 through 2024. The equal-jurisdiction population coefficient of variation fell from 15.62% to 12.11%, while the median state share rose from 0.335% to 1.263%. That is evidence of convergence in this particular published-claims measure — not proof that access, patient use, disease burden, or treatment rates became equal.

Medicare GLP-1 prescriptions by state: complete 2024 data

Medicare Part D recorded 21,825,833 claims nationally in the included analytic scope in 2024. The table orders all 50 states and D.C. by raw claims and adds a separate within-file comparison: included claims as a percentage of every published Part D claim attributed to prescribers in the same state.

CMS publishes the underlying geography-and-drug records, not this assembled table. The 51 jurisdictions account for 21,698,020 claims, or 99.414% of the national figure.

Source: The RX Index Research analysis of CMS Medicare Part D Prescribers — by Geography and Drug, data year 2024, file MUP_DPR_RY26_P04_V10_DY24_Geo.csv. Claims include original prescriptions and refills. State is the prescriber's practice location in NPPES. Standardized fills shown to one decimal. Gross drug cost includes payments by Part D plans, beneficiaries, government subsidies, and other third parties; does not reflect manufacturer rebates. Records from fewer than 11 claims are omitted by CMS. Dataset v1.1.0; last verified August 1, 2026.

Which states had the most Medicare GLP-1 prescription claims?

California led with 1,865,079 claims, followed by Texas with 1,811,427, New York with 1,473,609, and Florida with 1,448,671. Those four states accounted for 6,598,786 claims — 30.234% of the national total; the top ten accounted for 11,534,854 claims, or 52.850%.

Those are volume comparisons, not treatment-rate comparisons. The source file does not contain state Part D enrollment, beneficiary residence, diagnosis, or the demographic and plan-design variables needed to explain why one state generated more claims than another.

Two contrasts survive that limitation. Texas was second by raw claims and fifth by within-file share. California was first by raw claims and 31st by within-file share. Florida was fourth by raw claims and 48th by within-file share. Those statements describe the two measures in this dataset; they do not establish what caused the differences.

Source: The RX Index Research calculations from the 2024 CMS geography-and-drug file. The top-four and top-ten sums were recomputed from the published state table and reconciled to the national row.

Which states had the highest GLP-1 share of published Part D claims?

Alaska had the highest within-file share at 1.757%, followed by Oklahoma at 1.674% and Louisiana at 1.653%. Rhode Island was lowest at 1.025%, followed by Oregon at 1.053% and Montana at 1.105%.

This measure answers a narrow question: of every 100 published Part D claims attributed to prescribers in a state, how many fell inside the included analytic scope? It is not a beneficiary rate, per-capita rate, prevalence estimate, access measure, or treatment rate.

Source: The RX Index Research analysis of the 2024 CMS geography-and-drug file. "Share" means included claims divided by all published Part D claims attributed to prescribers in the same state. Records derived from fewer than 11 claims are absent from both numerator and denominator where applicable.

Why use this denominator?

The numerator and denominator come from the same file, year, and prescriber-geography definition. That avoids the specific mismatch created by dividing prescriber-location claims by beneficiary-residence enrollment.

It does not make the measure complete. CMS omits geography-drug records built on fewer than 11 claims, so the denominator is explicitly published Part D claims. Suppression can affect small states and low-volume products more than large cells. Alaska's first-place position persisted from 2020 through 2024, which shows it was not a one-year result, but the state had only 33,832 included claims in 2024 and should not be treated as interchangeable with a million-claim state.

Did the state spread widen or narrow from 2019 to 2024?

It narrowed in every annual file under the within-file measure. The highest-to-lowest ratio fell from 2.34× in 2019 to 1.71× in 2024, and the equal-jurisdiction population coefficient of variation fell from 15.62% to 12.11%.

Read the table as a whole and the pattern is unusually clean. The minimum, median, and maximum state shares all rose from 2019 through 2024 while the cross-state dispersion of that share declined.

Source: The RX Index Research analysis of the 2019–2024 CMS annual files. Each state and D.C. receives equal weight. Population coefficient of variation = population standard deviation of the 51 state shares ÷ their arithmetic mean × 100. Dataset v1.1.0.

The state positions also changed. North Carolina had the highest share in 2019 at 0.417%. Alaska moved to the top in 2020 and remained there through 2024. Oklahoma moved from 26th to second, Michigan from 37th to 17th, and Arkansas from 49th to 29th. Montana moved from 10th to 49th, Delaware from second to 25th, and North Dakota from 14th to 34th.

This is a descriptive convergence finding. The CMS file cannot determine whether it reflects formulary design, clinical adoption, disease burden, product supply, prescribing culture, cross-border care, or some combination of those factors.

How did Medicare GLP-1 claims change from 2019 to 2024?

Claims rose from 4,798,441 in 2019 to 21,825,833 in 2024 — an increase of 354.9%, or 4.55 times the 2019 volume. The five-interval compound annual growth rate was 35.4%.

Gross claim-level drug cost rose from $5.274 billion to $27.503 billion, or 5.21×. That cost field cannot be read as net Medicare spending because it combines amounts paid by several parties and does not reflect manufacturer rebates.

Source: The RX Index Research analysis of the 2019–2024 CMS geography-and-drug files. Gross drug cost is not attributable directly to Medicare Trust Fund payments and does not reflect manufacturer rebates. Dataset v1.1.0.

2023 had the largest annual increase in the series: 59.8%. The included brand mix also changed sharply that year: Ozempic became the leading brand in 39 of 51 jurisdictions, up from seven in 2022, while Mounjaro recorded its first full-year volume after entering the files in 2022.

Gross cost per claim peaked at $1,369.58 in 2022 and fell to $1,260.13 in 2024, an 8.0% decline. Cost per claim is a mix-sensitive measure: it changes with the products used, days supplied, dosage patterns, claim composition, and price components in the PDE data.

State raw-claim growth differed substantially. Rhode Island had the fastest 2019–2024 CAGR at 44.6%, from 10,606 claims to 67,017, followed by Arkansas at 42.4%, Oklahoma at 42.3%, Florida at 41.8%, and Colorado at 40.5%. The slowest were Montana at 28.1%, North Carolina at 30.5%, North Dakota at 30.6%, Delaware at 30.7%, and New York at 30.7%.

State claim totals, 2019–2024

The full six-year state history is below. It is ordered by 2024 claim count; the first column is not a 2019 ordering.

Source: The RX Index Research analysis of the 2019–2024 CMS geography-and-drug files. Claims include original prescriptions and refills and are assigned to the prescriber's NPPES practice location. Geography-drug records derived from fewer than 11 claims are omitted. CAGR uses five annual intervals. Dataset v1.1.0.

Which drugs accounted for Medicare Part D GLP-1 claims in 2024?

Ozempic was the leading included brand in all 50 states and D.C. in 2024. Nationally, it accounted for 10,413,554 claims, or 47.712% of the included total.

The leading-brand shift was not gradual at the state level: Trulicity led in 49 jurisdictions in 2019 and all 51 in 2020; Ozempic led in all 51 by 2024.

Source: The RX Index Research analysis of the 2019–2024 CMS geography-and-drug files. "Led" means the brand with the largest published included claim total within a state or D.C. for that year.

Source: The RX Index Research analysis of the 2024 CMS geography-and-drug file. Tirzepatide is included under the published methodology but is a GIP receptor and GLP-1 receptor agonist, not a GLP-1-only receptor agonist.

Four brands — Ozempic, Mounjaro, Trulicity, and Rybelsus — accounted for 96.545% of included 2024 claims. The remaining products accounted for 3.455%.

What changed for Wegovy and Zepbound in 2024?

Wegovy rose from 142 published claims in 2023 to 211,490 in 2024. The FDA approved Wegovy on March 8, 2024 to reduce the risk of cardiovascular death, heart attack, and stroke in adults with cardiovascular disease and either obesity or overweight. The timing is consistent with an indication becoming coverable under Part D, but the CMS file has no diagnosis field and cannot attribute the claims to that indication. FDA approval.

Zepbound first appeared in this series with 353 claims in 2024. The FDA approved it for moderate-to-severe obstructive sleep apnea in adults with obesity on December 20, 2024. The source file cannot establish why those claims were paid. FDA approval.

The brand value Liraglutide also first appeared in this panel in 2024, with 62,242 claims and gross cost of $772.54 per claim. The brand field alone does not establish dispensing-channel, manufacturer, or generic-market details beyond the value published by CMS.

State brand mix differed even though Ozempic led everywhere. Ozempic's share ranged from 54.7% in Iowa and North Dakota to 37.0% in Utah, a 17.8-point spread. Gross cost per claim ranged from $1,521.66 in Hawaii to $1,116.27 in Rhode Island, against a national average of $1,260.13.

What does this data show — and what does it not show?

It shows final-action Medicare Part D prescription-drug events, standardized 30-day fills, and gross drug cost for a fixed analytic scope, aggregated by prescriber practice location from 2019 through 2024. It supports descriptive comparisons of published claims, costs, brand mix, and changes over time.

It does not support patient-level, indication-level, residence-based, net-spending, or causal conclusions.

Claims are not unique patients

A claim is a dispensed prescription event, including original prescriptions and refills. The file includes a unique-beneficiary count for each individual geography-drug row, but adding those counts across products would double-count a beneficiary who used more than one included drug. A class-level unique-patient total cannot be recovered from the aggregate rows without beneficiary-level de-duplication.

State is not beneficiary residence

CMS assigns geography from the prescriber's practice location in NPPES. A prescriber in one jurisdiction can treat a beneficiary who lives in another. CMS also documents NPI-attribution issues, including claims submitted under organizational and individual NPIs and point-of-sale transcription errors.

The file does not identify why a drug was prescribed

There is no diagnosis or indication field. The data cannot separate type 2 diabetes, cardiovascular-risk reduction, obstructive sleep apnea, weight management, or other uses. This is not a dataset of Medicare weight-loss prescriptions.

Gross drug cost is not net Medicare spending

CMS defines total drug cost as ingredient cost, dispensing fees, sales tax, and applicable administration fees based on amounts paid by the Part D plan, beneficiary, government subsidies, and other third parties. CMS states that this PDE-derived amount is only one part of the payment process, cannot be directly attributed to the Medicare Trust Fund at the prescriber or drug level, and does not reflect manufacturer rebates. The public file cannot produce a defensible net-spending estimate.

These are filled claims, not prescriptions written and abandoned

The source is final-action prescription drug event data. It does not count a prescription that was written but never dispensed as a Part D event.

The scope is Part D only

Commercial insurance, Medicaid, cash-pay, compounded products outside Part D, Part B claims, and Medicare GLP-1 Bridge claims are outside this series.

The comparisons are descriptive

The file does not contain the demographic, clinical, formulary, supply, provider-market, or beneficiary-residence information needed to explain state differences causally.

How was this dataset built?

The RX Index downloaded six annual CMS geography-and-drug CSV files, verified the raw-file hashes, applied one fixed generic-name scope, aggregated the national and state records, built the within-file denominator, and reconciled the published geographies to CMS's national row. The complete package includes the source manifest, derived-file hashes, and a standard-library Python reproduction script.

The package and its headline outputs were reverified on August 1, 2026. The package SHA-256 matched the published value, all raw source hashes matched, and the principal derived CSV and JSON distributions reproduced byte-for-byte from the frozen source files.

Included generic names

The analytic scope includes every standalone GLP-1 receptor-agonist generic-name value observed in the 2019–2024 files, plus tirzepatide:

Tirzepatide is included because the methodology intentionally treats the policy-relevant incretin category together. The FDA identifies tirzepatide as a GIP receptor and GLP-1 receptor agonist. Current FDA Mounjaro label.

Albiglutide and lixisenatide are retained for a consistent six-year scope. Together, those two generic-name values account for 597 published claims across the complete panel.

Excluded combination products

The scope excludes Insulin Degludec/Liraglutide and Insulin Glargine/Lixisenatide. Both are fixed-ratio insulin/GLP-1 combinations; excluding them keeps the published series centered on standalone incretin products. The rule is explicit so the panel can be rebuilt under a broader definition without ambiguity.

Processing steps

Derived formulas

National reconciliation

Source: The RX Index Research analysis of the 2024 CMS geography-and-drug file. The 206-claim difference is consistent with the file's suppression structure, but the published data do not identify a geography for those claims.

Source files and hashes

The CMS catalog identifies the 2024 distribution as the latest complete annual file, recorded as modified May 21, 2026. The 2024 CSV is the RY26 release.

CMS field definitions used here

Primary definitions: CMS data dictionary and CMS methodological overview.

Why is there no per-beneficiary state rate?

A state claims-per-enrollee rate would divide a prescriber-location numerator by a beneficiary-residence denominator. Those are not the same geography, and the public aggregate files do not provide a defensible way to measure or correct the cross-border mismatch.

The within-file share answers a narrower question, but both sides of the fraction use the same prescriber geography, year, source, and suppression framework.

The source also contains Tot_Benes, a unique-beneficiary count for each individual drug row. It cannot simply be summed across the included products: a beneficiary who changed products or used more than one product would appear in more than one row. For that reason, this version does not publish state class-level user counts or claims per user.

Why can published Medicare GLP-1 totals differ?

Published totals can differ without either analysis being arithmetically wrong. The decisive questions are which CMS source was used, which generic names and combination products were included, which year or claim status was selected, and whether the cost figure is gross or rebate-adjusted.

The class-scope effect is large. In 2024, excluding tirzepatide would remove 5,103,909 claims from this dataset — 5,103,556 Mounjaro claims plus 353 Zepbound claims. A publication that says "GLP-1" but does not disclose whether tirzepatide is included cannot be reproduced precisely.

A peer-reviewed 2025 study in the Journal of the American Heart Association used proprietary Symphony Health data across payers and a BRFSS obesity denominator to examine 2023 state variation. That is a different population, source, denominator, and research question from this Medicare Part D prescriber-geography panel. The two should not be treated as interchangeable state rates. Study record.

Why is 2024 the current pre-Bridge baseline?

The 2024 CMS distribution is the newest complete annual geography-and-drug file available as of August 1, 2026, and every year in the 2019–2024 panel predates the Medicare GLP-1 Bridge. That makes it the current published baseline for ordinary Part D claims before the separate demonstration began — not the final calendar year before the Bridge.

The standard Part D definition incorporates the statutory treatment of agents when used for anorexia, weight loss, or weight gain. Products can still be coverable for medically accepted indications outside that excluded use, and CMS notes that individual plans may report otherwise excluded products through supplemental coverage. Social Security Act §1860D-2 and §1927(d)(2).

The Medicare GLP-1 Bridge began July 1, 2026 and runs through December 31, 2027. CMS says it operates outside the Part D benefit's coverage and payment flow, carries a $50 beneficiary copay, does not apply that copay to the Part D deductible or TrOOP, and does not provide LIS cost-sharing assistance for the copay. CMS Bridge overview.

As verified August 1, 2026, CMS lists Foundayo, Wegovy injection and tablets, and Zepbound KwikPen for eligible weight-management use; the agency says the product list may change. CMS information for providers.

Bridge claims and ordinary Part D PDE claims are different streams. Future versions of this panel should not merge them silently.

What are the limitations of the state data?

The state comparison is reproducible, but it is not a census of patients, diagnoses, residence, or net Medicare spending. The limits below determine what the figures can carry.

Small-cell suppression. CMS omits aggregated geography-drug records derived from fewer than 11 claims and can remove additional values for counter-suppression. Every state numerator and denominator is therefore based on published cells and can understate the true total.

Prescriber geography, not patient geography. The state is the provider practice location in NPPES. Cross-border care prevents a clean beneficiary-residence interpretation.

Claims are not patients. The class-level number of unique users cannot be obtained by summing drug-level beneficiary counts because that would double-count people who used multiple included products.

No diagnosis or indication. The file cannot identify weight-management claims or divide use among diabetes, cardiovascular, sleep-apnea, and other indications.

Gross cost, not net spending. The cost field includes several payers, cannot be attributed directly to Medicare Trust Fund payments at the drug level, and excludes manufacturer rebates.

Source exclusions and attribution issues. CMS excludes a small proportion of PDE records whose NDCs do not match the drug-information database and documents known NPI-attribution errors.

Class definition. Tirzepatide is included intentionally even though it activates both GIP and GLP-1 receptors; fixed-ratio insulin combinations are excluded. The exact list is published above.

Annual publication lag. The newest complete file covers 2024. The panel does not describe 2025 or 2026 Part D utilization and contains no Bridge claims.

Descriptive, not causal. State differences cannot establish effects of obesity prevalence, demographics, access, plan design, clinical practice, or supply without additional aligned data.

Frequently asked questions

Data downloads

The data behind every table are available in human-readable and machine-readable formats. All files return HTTP 200.

Package SHA-256: c55dfc1d6a1ccf1a9eafa4f5e844778434694d52640e38d9d0b6ac1bf6bc9c61

Primary sources: CMS dataset · CMS data dictionary · CMS methodology · Social Security Act §1860D-2 · CMS Medicare GLP-1 Bridge

How to cite this page

The RX Index Editorial Team. "Medicare GLP-1 Prescriptions by State:
2024 Part D Claims Data." The RX Index Research. Dataset v1.1.0.
Last verified August 1, 2026.
https://therxindex.com/research/medicare-glp-1-prescriptions-by-state/
Accessed [Month Day, Year].

The RX Index Research analysis of CMS Medicare Part D Prescribers
by Geography and Drug, data year 2024. Dataset v1.1.0, assembled
July 31, 2026, last verified August 1, 2026.
https://therxindex.com/research/medicare-glp-1-prescriptions-by-state/

Each table includes its data year, source, geography definition, cost qualifier, and suppression note directly beneath the table.

Version history

Related research

The RX Index Research is an independent research and reference resource tracking prices and access for prescription medications. This page is educational and is not medical, financial, or legal advice. Coverage questions about an individual Medicare plan should go to the plan or 1-800-MEDICARE; clinical questions belong with a licensed clinician. This page contains no sponsored content and no commercial relationship influenced the findings.