The Medicare Spending Story
A guided read through public Medicare physician/supplier payment data, Medicare enrollment, industry payments, and the federal exclusion list — written for data and analytics readers. Each chapter states a pattern, the modeling choice behind it, and the caveat that limits it. Every number below is computed from the published data marts when the site is built; nothing is typed in by hand. Patterns are anomalies and leads for further review, not accusations or findings about any state, provider, or company.
Sources: CMS Medicare Physician & Other Practitioners public use file (PUF), CMS Medicare Monthly Enrollment, CMS Open Payments (General payments), and the HHS-OIG List of Excluded Individuals/Entities (LEIE). Public aggregates only — no patient-level data. Not medical, legal, or billing advice.
1. The headline: payments grew, and faster in some states
National growth, first to latest year
California, same period
Minnesota, same period
Medicare physician/supplier payments in the PUF rose from $89.0B in 2022 to $97.3B in 2024 , a 9.3% increase; the latest year alone added $3.6B ( 3.8% ). Payment here is the PUF's average Medicare payment per service times the number of services, summed to state × year.
The two worked examples sit at different points of the distribution. California grew 20.2% over the period to $12.8B , with 11.4% in the latest year alone — rank 2 of 26 states with at least $1B in first-year payments (the fastest was AZ at 23.1% ). Minnesota, just under that size line at $0.99B in the first year, grew 6.0% to $1.05B ( 4.0% in the latest year) — below the national rate. Raw growth says nothing yet about why. → Medicare spend YoY by state
2. Is it just more beneficiaries?
Pearson r: FFS Part B enrollment vs spend growth
FFS Part B enrollment change
Total Medicare enrollment change
The PUF covers fee-for-service (Original Medicare) claims, so the right denominator is fee-for-service Part B enrollment, not all Medicare beneficiaries. Over 2022→2024 , total Medicare enrollment grew 4.5% , but fee-for-service Part B enrollment changed -5.6% as people moved into Medicare Advantage, and it fell in 50 of 51 jurisdictions. Spending still rose 9.3% , so spend per fee-for-service enrollee rose faster than raw spend.
Across states, enrollment does explain part of the pattern: the correlation between fee-for-service enrollment growth and spending growth is r = 0.77 (about 59% of variance), versus r = 0.50 against total enrollment — the choice of denominator matters. In California, spend per fee-for-service enrollee grew 21.2% (to $4,545 ) with a Medicare Advantage share of 51% ; in Minnesota, 13.5% (to $2,739 ) with 59% in Medicare Advantage. Correlation is not causation; case mix, prices, and site-of-care shifts also move these ratios. → Enrollment vs spending growth
3. Where the growth concentrates
Added by the top 10 specialties
Added by the top 20 HCPCS codes
Net national increase
Growth is not spread evenly. In the latest year, the ten specialties (of 113 ) with the largest dollar increases added $3.74B — 105% of the net national increase of $3.57B . A share above 100% is possible because other specialties shrank ( 39 declined), offsetting part of the growth. The largest single contributor was Nurse Practitioner (up $1.05B ). Specialty labels are self-reported billing categories, so shifts between them can reflect who bills a service rather than more of it. Percentage growth is noisy on small bases, so rankings by percent use a volume floor (at least $5M prior-year payment and 100 billing NPIs): under that floor the fastest-growing specialty was Preventive Medicine at 64.9% .
By billing code, the twenty HCPCS codes (of 6,471 ) with the largest dollar increases added $5.31B , or 149% of the net national increase. As with specialties, a share above 100% means growth among the top gainers is partly offset by declines elsewhere ( 3,484 codes declined). Among codes with at least $1M in prior-year payments, the largest increase was Q4205 ( Membrane graft or membrane wrap, per square centimeter ), up $552.2M . Code-level jumps often track coverage, pricing, or coding changes, so they are where review starts, not where it ends. → Specialty & HCPCS concentration
4. Who drives it — benign explanations first
Share of net increase from top 500 NPIs
…of which organizations
In 2024 , the 500 billing NPIs with the largest year-over-year payment increases added $3.11B — 87.2% of the national net increase of $3.57B . Before reading anything into that, the ordinary explanations come first: 153 of the 500 are organizations, where practice acquisitions, mergers, and consolidation of billing under one NPI produce large jumps with no change in care delivered; new service lines, new drugs billed under Part B, and payment-rate updates do the same. A large share of a net figure also partly reflects declines elsewhere. 0 of these top-ranked NPIs match the exclusion list. → High-growth NPI concentration
5. The exclusion-list overlay — timing matters
LEIE-listed NPIs in Medicare payment data
…excluded in a later year
LEIE records that carry an NPI
Joining the HHS-OIG exclusion list to the PUF on NPI finds 443 NPIs across 846 provider-years from 2022 to 2024 , with $118.4M in payments. Timing is the first split: 434 NPIs ( $114.0M ) were excluded in a year after the payments shown, which is expected — exclusions follow investigations of earlier conduct. 27 appear in their exclusion year ( $1.9M ), where services before the exclusion date are normal. Only 2 appear in a service year after their exclusion year ( $2.5M ); reinstatement, waivers, data lags, and NPI reuse need checking before any conclusion.
This is a floor, not a census: only 10.6% of the 84,093 LEIE records carry an NPI, and the join uses NPI only (no name matching). → LEIE overlay
6. Industry money — transparency context
LEIE-listed NPIs with reported industry payments
Reported general payments
…also in Medicare payment data
Open Payments is a disclosure program: manufacturers report what they pay clinicians. In program years 2023 – 2024 , manufacturers reported 11,274 general payments worth $5.16M to 638 NPIs that appear on the exclusion list; 222 of them also appear in the Medicare overlay in chapter 5. 284 NPIs received payments ( $511,796 ) in a program year after their exclusion year. Exclusion bars billing federal health programs; it does not by itself bar receiving industry payments, so this chapter is transparency context, not a billing finding. → Open Payments × LEIE
7. Method and caveats, in plain language
- Pipeline. Public files are stored as-is, cleaned in a staging layer, and modeled into tested data marts; every chart and number on this site reads only from those marts.
- No county detail. The physician PUF reports provider location, not where patients live, so there is no county view of spending.
- No beneficiary-level data. Everything is aggregated by CMS before release; nothing here describes an individual patient.
- Small cells are suppressed. CMS withholds rows with very few beneficiaries, so small providers and rare services are undercounted, and percent changes on small bases are unstable — hence the volume floors in chapter 3.
- NPI-only joins. Exclusion-list and Open Payments links use NPI alone, which misses records without an NPI (chapter 5) and can't detect identity changes.
- Leads, not findings. Each pattern is a starting point for review alongside context the data can't show.
Start from the Key Findings summary for the headline numbers, or open any chapter's page for tables and charts.
Full lineage, model grain, test inventory, and validation status: Methods.