How the numbers are built
Every price on this site comes from a public file a hospital or CMS published. This page shows the path from those files to a number on a page, and what we do (and don't do) along the way.
Sources
| What | Where it comes from | Vintage |
|---|---|---|
| Facility prices (cash, negotiated) | Each hospital's CMS machine-readable file, linked from its cms-hpt.txt or price-transparency page | Whatever the hospital last published; re-fetched weekly |
| Hospital outpatient benchmark | Medicare OPPS Addendum B, national unadjusted payment rate | CY2025 |
| Physician / office benchmark | Medicare PFS: non-facility total RVU × conversion factor ($32.3465) | CY2025 |
| Lab benchmark | Medicare Clinical Laboratory Fee Schedule | CY2025 |
| Inpatient (MS-DRG) benchmark | IPPS: DRG relative weight × (operating + capital national base rate), wage index 1.0 | FY2026 Final Rule Tables 5 and 1A/1D |
| ASC benchmark | Medicare ASC Addendum AA/BB national payment rate; packaged codes shown as not separately payable | January 2026 update |
| Quality | Care Compare overall star rating; hip/knee replacement complication rate (COMP_HIP_KNEE), matched to facilities by CCN or name | Latest Care Compare release |
What we compute, exactly
Which rows are kept
Only rows whose code is on the tracked list (82 codes). A hospital file typically has hundreds of thousands of rows; a few thousand survive. Rows without a usable code, and files that contain only chargemaster items with no CPT/HCPCS/DRG codes, are dropped and logged.
The per-facility price
For each facility and code: cash is the published discounted cash price; negotiated min / median / max summarize every payer row for that code. The facility list shows cash where published, otherwise the negotiated median.
Payer rates
Raw plan names are collapsed to 8 insurers (all Aetna plans → Aetna, Blue Cross / Blue Shield / Anthem → BCBS, and so on). The rate shown for a facility and insurer is the median across that insurer's plans. Plans that don't map to a major insurer are excluded from the payer view but still count in the negotiated range.
Typical price and the band
Across the facilities in the selected area, the typical price is the median with implausible values excluded: anything under 25% of the Medicare rate (a mis-keyed file), and anything outside the Tukey fences (1.5× the interquartile range). The band shown is the 25th to 75th percentile; flagged outliers stay visible in the table, marked.
Medicare multiples and the ask
Every price is expressed as a multiple of the Medicare benchmark for its setting (OPPS for hospital outpatient, IPPS for inpatient DRGs, PFS for office). The suggested ask is the lower of 2× Medicare or the local cash price, a figure most facilities accept for self-pay.
Site of care
Medicare's own hospital-outpatient (OPPS) and ASC rates for the same code give the reference differential. The per-case shift figure is the hospital typical price minus the median of published non-hospital prices for that code, using only sites that publish a price.
Percentiles
"Your rate is at the Nth percentile" means N% of facilities in the selected column have a lower published rate. It is computed over the facilities shown, not a national sample.
What this is not
- Not a quote. Files carry facility fees; professional (surgeon, anesthesia) fees are often separate. Some hospitals publish gross charges only, or one price for a bundle of codes.
- Not a contract. A payer column is a facility's median across that insurer's plans, and hospitals sometimes publish algorithms instead of dollars. Verify before relying on a number.
- Not complete. Ambulatory surgery centers and physician offices are not required to publish files; the non-hospital rows here are the ones that chose to. Quality measures are blank where CMS reports too few cases.
- Not advice. Educational estimates from public data; not legal, medical, or financial advice, and not a guarantee of what anyone will pay.
Freshness and reproducibility
The whole dataset is rebuilt from scratch every week: every file is downloaded again, parsed, filtered, and swapped in atomically, so the site never serves a half-built dataset. Hospital files change on their own schedules; the vintage shown on each page is our last rebuild (September 2026). Every source above is public and linkable, so any number here can be traced back to a row in a hospital's own file.