Medicare GLP-1 Prescriptions by State, 2019–2024
Version 1.1.0
Assembled and last verified: July 31, 2026

PURPOSE
This is a reproducible state-level aggregation of the Centers for Medicare & Medicaid Services (CMS)
Medicare Part D Prescribers - by Geography and Drug annual files. It answers the literal search
question "Medicare GLP-1 prescriptions by state" while preserving the distinction between paid
prescription claims, 30-day-equivalent fills, unique beneficiaries, and patient residence.

CORE DRUG SCOPE
Included CMS generic-name values:
- Albiglutide
- Dulaglutide
- Exenatide
- Exenatide Microspheres
- Liraglutide
- Lixisenatide
- Semaglutide
- Tirzepatide

The scope includes every standalone GLP-1 receptor-agonist generic name observed in the 2019–2024
CMS files. Tirzepatide is included because FDA identifies it as a dual GIP/GLP-1 receptor agonist and
it is routinely grouped with GLP-1 drugs in public-policy discussion.

Excluded fixed-ratio insulin/GLP-1 combinations:
- Insulin Degludec/Liraglutide
- Insulin Glargine/Lixisenatide

GEOGRAPHY
The published 51-jurisdiction tables include the 50 states and District of Columbia. CMS state
geography is the prescriber's practice location recorded in NPPES, not the beneficiary's residence.
National CMS totals also include claims assigned to Puerto Rico, U.S. territories, armed-forces
geographies, foreign locations, and unknown geography.

KEY DEFINITIONS
- Claim: CMS Tot_Clms; a paid Part D prescription claim, including original prescriptions and refills.
- 30-day fill: CMS Tot_30day_Fills; standardized 30-day-equivalent fills.
- Gross drug cost: CMS Tot_Drug_Cst; ingredient cost, dispensing fees, sales tax, and applicable
  administration fees paid by Part D plans, beneficiaries, government subsidies, and other payers.
  It is not net of confidential rebates or manufacturer concessions.
- Published all-Part-D claims: sum of all published state-level drug rows in the CMS geography file.
  CMS omits aggregated geography-drug records with fewer than 11 claims, so this denominator is not
  a mathematically complete count of every Part D claim.
- State GLP-1 share: published state GLP-1 claims divided by published all-Part-D claims in that state.
  It is not a per-beneficiary, per-capita, prevalence, or patient-residence rate.

METHOD
1. Download the annual CMS geography-and-drug CSV for each year from 2019 through 2024.
2. Verify every frozen source file against its SHA-256 hash.
3. Keep National and State geography rows.
4. Include rows whose Gnrc_Name exactly matches one of the eight included generic names.
5. Exclude fixed-ratio insulin/GLP-1 combinations by not including their generic-name values.
6. Sum Tot_Clms, Tot_30day_Fills, and Tot_Drug_Cst by year and prescriber geography.
7. Sum every published drug row inside each state to create the published Part D denominator.
8. Group filtered rows by brand inside each state and select the largest claim count.
9. Rank raw claims and GLP-1 share independently.
10. Reconcile the 51-jurisdiction sum to the CMS national total and disclose the residual.

Equivalent class filter:
    Gnrc_Name IN (
      'Albiglutide',
      'Dulaglutide',
      'Exenatide',
      'Exenatide Microspheres',
      'Liraglutide',
      'Lixisenatide',
      'Semaglutide',
      'Tirzepatide'
    )

HEADLINE RESULTS
- 2019 CMS national total: 4,798,441 claims
- 2024 CMS national total: 21,825,833 claims
- 2024 50 states plus D.C.: 21,698,020 published claims
- 2024 identified claims outside the 50 states plus D.C.: 127,607
- 2024 national-minus-all-published-state-geography residual: 206 claims
- California: 1,865,079 claims (largest raw state count in 2024)
- Alaska: 1.757179% of published Part D claims (largest share in 2024)
- Ozempic was the largest included brand by claim count in all 50 states and D.C. in 2024.

IMPORTANT LIMITATIONS
- Claims are not unique patients.
- CMS omits aggregated geography-drug records with fewer than 11 claims. This creates small
  undercounts for low-volume products and makes the published all-Part-D denominator incomplete.
- The file does not identify diagnosis or indication.
- State means prescriber location, not patient residence.
- Gross drug cost is not net spending.
- These are paid Part D claims, not all prescriptions written and not all GLP-1 use.
- The Medicare GLP-1 Bridge began in 2026 and operates outside the Part D claim flow represented here.
- Do not interpret geographic differences as causal effects without additional demographic,
  eligibility, plan-design, clinical, supply, and prescriber-market data.

PRIMARY DOCUMENTATION
CMS dataset catalog: https://catalog.data.gov/dataset/medicare-part-d-prescribers-by-geography-and-drug
CMS data dictionary: https://data.cms.gov/sites/default/files/2026-05/3e80b44e-5a87-4414-959a-b6a7c8c18720/MUP_DPR_RY26_P04_V10_DY24_Geo.pdf
FDA tirzepatide classification: https://www.accessdata.fda.gov/drugsatfda_docs/label/2026/215866s009lbl.pdf
FDA albiglutide classification: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/125431s019lbl.pdf
FDA lixisenatide classification: https://www.accessdata.fda.gov/drugsatfda_docs/label/2024/208471s009lbl.pdf

SOURCE FILES AND SHA-256

2019: https://data.cms.gov/sites/default/files/2021-08/MUP_DPR_RY21_P04_V10_DY19_Geo.csv
SHA-256: d87977e5aed2ba7c862165bba0a32e844823ecd1f8cbeebe52a8bcd4f1892e01

2020: https://data.cms.gov/sites/default/files/2022-07/ca71b7df-4d48-4c2d-aded-2ca22285739c/MUP_DPR_RY22_P04_V10_DY20_Geo.csv
SHA-256: 28a7f5558435ac99cb98b113ad054e54cb4fa487b623ae55dc89f6b152a9f57b

2021: https://data.cms.gov/sites/default/files/2023-04/3d3ebd5b-b4bf-45b4-876d-afa7916d1b72/MUP_DPR_RY23_P04_V10_DY21_Geo.csv
SHA-256: e345008b016df5d1e7362f82a0139a57f2d42ceae8f37c3ddc981168c9286535

2022: https://data.cms.gov/sites/default/files/2024-05/410ed206-d782-4dba-b442-6bcd45ae2016/MUP_DPR_RY24_P04_V10_DY22_Geo.csv
SHA-256: a26588a60a1e6dbb1d39b83d750ed68029b76d35bb08bd72a418172eca867f87

2023: https://data.cms.gov/sites/default/files/2025-04/9fe6b8a6-0cb9-4b7c-9760-87800da010a8/MUP_DPR_RY25_P04_V10_DY23_Geo.csv
SHA-256: 876ff262c8a3eb0b8fca312f2004f14d42a8bfd01829e0695d64e34b5191f4f8

2024: https://data.cms.gov/sites/default/files/2026-05/3e80b44e-5a87-4414-959a-b6a7c8c18720/MUP_DPR_RY26_P04_V10_DY24_Geo.csv
SHA-256: c84a834ffb34fa1e46c6b8566d1e89d87f40390120393342e050f7203ef3ed68

DERIVED DATA FILES AND SHA-256

medicare_glp1_state_panel_2019_2024.csv
SHA-256: 6d077d8a575126428e820723cd392c9bd7230e79994c36edf5e8b289f18accbc

medicare_glp1_state_panel_2019_2024.json
SHA-256: c584c0df7d165ea9ae37f9e160d0f9fa505c11354350fbca7dba11dbdd1a9b58

medicare_glp1_state_trends_2019_2024.csv
SHA-256: 00c10929a2c7ad3167571801f987d58ca0376b0ac91d4e4214f6cdd72858fabe

medicare_glp1_national_trend_2019_2024.csv
SHA-256: 47a7a84a18330a438ecba8d013f19fc9b500a4c20b2da9184af13463211c6a99

medicare_glp1_national_brand_mix_2019_2024.csv
SHA-256: 70fabf236fec0cd3a3e0e26c829467bf96611aa6ca681f54e37f18906085bf82

medicare_glp1_prescriptions_by_state_2024.csv
SHA-256: 6b17bd2fcaf6a755516a6301a0a40bda8841a4e26307ef12826b3ea6eb2a0a22

medicare_glp1_prescriptions_by_state_2024.json
SHA-256: 9a45b1facee0136a1b83fc8ba8254a036e95008a23185206f6240c28efb0d4d4

REPRODUCTION
Run:
    python build_medicare_glp1_dataset.py --source-dir /path/to/cms/files --output-dir /path/to/output

Without --source-dir, the script downloads the frozen CMS files and verifies their hashes.
