HomeData sources

Where our data comes from

Even doesn't generate prices. We surface what hospitals already publish under federal law, plus a few public datasets that make it usable. Everything below is either open public data or used with the source's permission.

1,638 hospitals indexed·7.5M price rows·Updated April 2026
01

CMS Hospital Price Transparency files

Source
What it is

Every hospital's machine-readable price file: chargemaster line items, shoppable services, discounted self-pay rates, and (for some hospitals) payer-specific negotiated rates (Aetna, BCBS, Cigna, Humana, UHC, Kaiser). Published by hospitals under a federal rule that took effect January 1, 2021.

Why we use it

This is the primary source for everything on Even. Each hospital publishes their own file at a URL we ingest. Hospitals are required by federal law (45 CFR Part 180) to keep these public and machine-readable.

License

Public data — hospitals are required to publish it. No license restrictions on consuming or republishing.

02

CMS Hospital Compare / Care Compare

Source
What it is

The list of US hospitals: facility names, addresses, phone numbers, ZIP codes, ownership type, plus CMS quality star ratings (1-5).

Why we use it

How we know which hospitals exist and how to match published prices to a real facility. Star ratings let us show a quality signal alongside price.

License

US federal government work — public domain.

03

CMS Medicare Physician Fee Schedule

Source
What it is

What Medicare pays for ~7,500 CPT-coded procedures. We use this as the "fair price" benchmark next to each hospital's self-pay price.

Why we use it

Medicare's rate is the closest thing to an objective floor on what a procedure "should" cost. A hospital charging 30× Medicare for an MRI is a red flag.

License

US federal government work — public domain.

04

CPT® codes (American Medical Association)

Source
What it is

The 5-digit procedure codes (like 72148 for a lumbar-spine MRI) that hospitals, insurers, and Medicare all use to identify services.

Why we use it

CPT codes are the shared key that lets us match a hospital's published price to the Medicare benchmark for the same service. We display the numeric codes with our own plain-English procedure names.

License

CPT® is a registered trademark of the American Medical Association. Codes shown for identification; descriptions on Even are our own consumer-friendly names.

05

GeoNames postal codes (US)

Source
What it is

Latitude/longitude centroids for every US ZIP code (~42,000 ZIPs).

Why we use it

We use these to (a) plot hospitals on the map and (b) sort search results by distance from the user's ZIP.

License

Creative Commons Attribution 4.0.

06

Mapbox

Source
What it is

The interactive map renderer (tiles, streets, building outlines, pan/zoom).

Why we use it

Visualizing where hospitals are relative to the user. Mapbox is the same renderer most major consumer apps use (NYT, Foursquare, Strava, etc.).

License

Commercial use under Mapbox terms.

Methodology

How we put it together

Hospital matching

The two main data sources (CMS Hospital Compare and CMS price transparency files) use different naming conventions for the same hospital — for example, ‘ST. MARY'S HOSPITAL’ on one side and ‘saint marys hospital’ on the other. We compute a canonical name for both, then JOIN on that. Where names diverge fundamentally (different abbreviations, system rebrands), we fall back to fuzzy matching or skip the row rather than guess.

Price ranking

When you search a procedure with a ZIP code, we compute the distance from your ZIP centroid to each hospital and keep hospitals within your chosen radius (50 miles by default). Within that radius, results sort by lowest price — you can switch to sorting by distance or CMS quality rating. If nothing falls inside the radius, we widen it automatically instead of showing an empty page.

Price-source preference

For each (hospital, procedure) pair, we prefer prices in this order: (1) the insurer-specific negotiated rate if you've picked an insurer and the hospital published it, (2) the discounted self-pay / shoppable bundle price, (3) the chargemaster ‘list price’ only as a last resort. We label which type of price you're seeing so you know whether it's a realistic out-of-pocket estimate or an aspirational sticker number.

Anomaly filtering

Some hospital transparency files have parser-broken rows ($0.01 placeholders, $999,999,999 sentinels, cash-price columns that got swapped with chargemaster). We filter rows where the cash price is higher than the gross charge (impossible by CMS definition), where the gross is less than $5, or where the gross is more than $500,000 (no single CPT line item legitimately costs that much). About 0.4% of rows get filtered.

What we don't do

We don't generate prices. We don't predict prices. We don't negotiate with hospitals. We don't have inside information. Everything on this site is a presentation of data that hospitals were already required to publish — usually in a form so unreadable that almost no patient sees it. Our job is the readable form.

Public data, made useful

Hospital prices have been federally required to be public since 2021. Almost nobody has been able to actually read them. Even is a thin layer on top of that public data — same numbers the hospitals already published, just in a place a person can actually use.