Open data

Open Nursing Home Data

Every month we rebuild a facility-level dataset of U.S. nursing homes from the public files CMS publishes, tag every column with its source, validate it against CMS's own totals, and publish it with checksums so anyone can retrieve the same files and get the same numbers.

Release listing temporarily unavailable
CMS values public domain · derived columns CC BY 4.0

The data store did not answer just now, so the download table cannot be shown. The build code, codebook and methodology are always available in thepublic repository; try this page again in a few minutes.

What is in a release

A release is one CMS vintage, named after the year and month of the Care Compare processing date. It joins seven CMS files (Care Compare provider information, penalties and ownership; PECOS enrollments, owners and changes of ownership; the chain performance file) on the CMS Certification Number. Columns tagged cms are CMS values cast to a type and nothing else; columns tagged tcr are documented derivations such as the owner-versus-party disclosure flags, the change-of-ownership counts, the rescinded staffing-minimum tests and the state Care Quality Index.

Reproducible
The build is a small open-source Python package. Given the same CMS files, it produces the same bytes; every release records the source files' checksums and vintages.
Validated
Every release is reconciled with the national row of the CMS chain performance file (facility counts, Special Focus Facilities, abuse icons, fines, payment denials) before it is published.
Versioned
Releases are never overwritten. Cite the release name and readers can retrieve exactly the files you used.
Honest about ownership
The ownership flags are the facility's own answers on Form CMS-855A. A facility that does not disclose a private-equity or REIT owner has not reported one; that is not a finding that it has none.

The history store

Releases are latest-only. The history lives beside them, rebuilt from public archives: every monthly snapshot of the CMS nursing-home files since January 2019, harmonized across CMS's three column-naming eras, plus the Special Focus Facility lists recovered from Internet Archive captures of the CMS PDF.

Facility history
One row per nursing home per monthly CMS snapshot: ratings, staffing, penalties, Special Focus status, chain, case-mix hours and the cycle-1 survey score.
Penalty history
Every distinct fine and payment denial seen in any snapshot, with the first and last snapshot it appeared in.
Ownership history
Every Care Compare owner relationship with first and last seen dates.
Special Focus Facility history
Every table of every archived edition of the CMS Special Focus Facility list, with CCNs matched by name, state and ZIP where the PDF printed none.

Per-snapshot raw and harmonized files, the coverage table and the manifests are listed in history/manifest.json andhistory/sff_manifest.json. Sections 9 and 10 of the methodology describe the column eras and the PDF layouts.

Licensing at a glance

CMS source values are U.S. government work in the public domain and need no attribution. Our derived columns, the arrangement of each release and the documentation are CC BY 4.0. The identifier crosswalk is CC0 so it can feed public knowledge bases. The build code is MIT. Read the full license notice.

Cite this data

Cite the release name you used and this page: citation formats.

All releases

The release list could not be loaded.

How it is built, and how to report a problem

The build pipeline, schema, tests and documentation are public at github.com/ttiatun/careratings-open-data. A release is built the month CMS publishes a new Care Compare vintage, validated, and published with its manifest; the changelog records every change to the format or the derivations. Our editorial standards describe what we will and will not claim from this data, and the corrections log records every correction to a published release or report.

Found an error, or need a column we do not publish? Write to our contact page and name the release and table. Facilities that want to correct their own record should contact CMS, which is the source of every value here.