Go beyond the headlines. These analyses dig into specific aspects of Medicare spending using 96 million rows of CMS physician payment data spanning 11 years (2014-2024).
Office vs Facility: Where Medicare Money Flows
Spending Hotspots by City & Zip Code
The Geographic Cost Gap
Charged vs Allowed vs Paid: The Three-Way Gap
Individual Doctors vs Corporate Medicine
What Doctors Charge vs What Medicare Pays
Medicare's Pharmaceutical Pipeline
The Geographic Divide in Medicare
10 Years of Medicare Spending
| Dimension | Coverage |
|---|---|
| Time Period | 2014-2024 (11 years) |
| Total Payments | $940B+ |
| Unique Providers | 1.82 million |
| Total Rows | 96 million+ |
| States & Territories | 61 |
| Medical Specialties | 105+ |
| Procedure Codes (HCPCS) | 7,500+ |
| Deep-Profiled Providers | 30,000+ |
| AI-Flagged for Fraud | 500 |
| Investigation Articles | 74+ |
Every analysis on OpenMedicare starts with the same foundation: CMS Medicare Provider Utilization and Payment Data, covering 11 years from 2014 through 2024. This dataset includes every payment Medicare Part B made to every physician and supplier in the United States — over 96 million individual rows of billing data.
We process this data to calculate aggregate statistics at the provider, specialty, state, and procedure level. Our analyses use standard statistical methods including peer comparison (comparing providers against specialty averages), trend analysis (year-over-year changes), geographic normalization (adjusting for cost-of-living differences), and outlier detection (identifying statistically unusual billing patterns).
For fraud-related analyses, we employ machine learning models trained on the billing patterns of 2,198 confirmed Medicare fraudsters. Our current model (v2) achieves an AUC of 0.83, meaning it correctly distinguishes between fraudulent and legitimate billing patterns 83% of the time.
We don't sample — we analyze every row. All 1.82 million providers, all 96 million+ billing records, across all 50 states and territories.
Most Medicare analyses look at a single year. Ours spans 2014-2024, revealing trends that only emerge over time — like the slow rise of telehealth or the steady concentration of spending.
Machine learning models don't just flag outliers — they identify patterns across thousands of variables that human auditors would never catch.
Every number on this site comes from publicly available CMS data. We show our sources, explain our methods, and let you verify everything yourself.
Physician-administered drugs grew from 11% to nearly 15% of Medicare spending in a decade. A single eye injection drug costs more than most federal agencies' budgets.
Providers submitted $3.5T+ in charges but Medicare paid $940B. The 3.7x average markup ratio has been climbing every year since 2014.
Rural providers earn less per service, treat fewer patients, and face hospital closures — while their patients are older and sicker on average.
Our AI flagged 500 providers whose billing patterns match convicted fraudsters — many have been billing Medicare for years without scrutiny.
Large healthcare organizations receive an increasing share of Medicare payments, while solo practitioners decline. The top 1% of providers collect over 25% of all payments.
New here? Start with the Spending Trends page for the big picture — how Medicare spending has changed over 11 years.
Interested in fraud? Head to the Fraud Hub to see which providers our AI flagged and why.
Curious about your state? Browse State Pages for detailed breakdowns of Medicare spending in your state.
Want to look up a doctor? Use the Provider Lookup to search any of 1.82 million Medicare providers.
Checking costs? Our Cost Calculator estimates Medicare payments for any procedure combination.
Part B only: Our data covers Medicare Part B (physician/supplier) payments. It does not include Part A (hospital inpatient), Part C (Medicare Advantage plan-level payments), or Part D (pharmacy) spending. Total Medicare spending across all parts exceeds $890 billion annually.
Fee-for-service: As more beneficiaries enroll in Medicare Advantage (now over 50%), our fee-for-service dataset represents a shrinking share of total Medicare beneficiaries. MA plan payments to providers are not included in CMS public use files.
Privacy thresholds: CMS suppresses data for providers with fewer than 11 beneficiaries for any given service to protect patient privacy. This means some low-volume providers and rare procedures are underrepresented in our analysis.
Note: All data is from publicly available Medicare records. OpenMedicare is an independent journalism project not affiliated with CMS.