How CMS Care Compare Nursing Home Data Flows: Source Datasets, Refresh Cadence, and What Gets Published vs. Withheld
CMS Care Compare — the federal consumer-facing database of nursing home quality information — is not a single dataset. It is a composite of at least six distinct data feeds, each with its own collection mechanism, publication delay, and suppression rules. Survey deficiency data, payroll staffing records, resident assessment quality measures, cost report financials, ownership disclosures, and enforcement actions all reach Care Compare on different timelines and through different CMS systems. Understanding the architecture of Care Compare is essential for interpreting what it shows and, equally important, what it withholds.
Architecture overview: six data feeds
Care Compare's nursing home section aggregates six primary data feeds: (1) standard and complaint survey deficiencies from CASPER/ASPEN; (2) payroll-based staffing from PBJ; (3) resident assessment quality measures from the MDS 3.0 warehouse; (4) Medicare claims quality measures from the Medicare claims data warehouse; (5) ownership and organization data from PECOS and CMS-2540 cost reports; and (6) enforcement action records from CMS regional offices and CMS's Integrated Survey and Certification Online (iQIES) system.
Each feed operates on its own collection and publication schedule. The resulting Care Compare display is a mosaic of data from different points in time — a facility's health inspection rating may reflect a survey from 11 months ago, while its staffing rating reflects PBJ data from 3 months ago. This temporal mismatch is a structural feature of the system, not a bug, but it means a snapshot of Care Compare does not represent a single point in time.
CMS publishes the underlying datasets through the CMS Data Portal (data.cms.gov). The provider data is released monthly via the Care Compare data archive, and the PBJ staffing data is released quarterly as a flat-file download. Fonteum ingests these releases as part of its standard source-family update cycle.
Survey and deficiency data
Survey deficiencies are the most consequential data element in Care Compare from a consumer standpoint. When a state survey team cites a deficiency, the surveyor records it in the ASPEN survey management system on a portable laptop during the survey. The completed CMS-2567 is uploaded to CASPER after the survey, where it goes through a supervisory review process before being accepted.
Once accepted, the deficiency data propagates to the Care Compare backend on the monthly refresh cycle. The data appears under the 'Health Inspections' tab with the actual F-Tag number, scope and severity, inspection date, and the full text of the CMS-2567 deficiency statement. This level of granularity is unusual in federal consumer disclosure databases — most publish summary statistics, not primary source records.
Survey records are retained on Care Compare for 36 months. Records older than 36 months roll off the public display but remain in the underlying CASPER database and are available for research through CMS data requests. The 36-month window is the basis for the health inspection star rating calculation.
Staffing data: PBJ pipeline and delays
PBJ submission requires facilities to upload a structured XML or CSV file with daily employee-level records covering all paid hours (not just hours worked at the bedside), employee type codes (RN, LPN, CNA, therapy, etc.), and daily census counts. The submission is due 45 days after quarter-end, with a correction period extending an additional 30 days.
After the correction window closes, CMS runs the staffing rate calculation pipeline. This pipeline applies the case-mix adjustment using MDS-sourced expected HPRD, then scores facilities against the national cutpoint table. Updated staffing ratings appear on Care Compare approximately 4–6 weeks after the correction window closes — meaning staffing data on Care Compare can be 3–4 months old at any point in the quarterly cycle.
CMS releases PBJ daily staffing data as quarterly flat files on data.cms.gov. The facility-level aggregates used in Care Compare are calculated by CMS from these same records. Fonteum's staffing deserts study used the raw daily PBJ records (1.32 million rows, 2025Q2) to compute county-level staffing availability, which is a different aggregation than what appears on Care Compare.
Quality measures: MDS and claims pipelines
Quality measures are calculated from two distinct data streams. Long-stay measures use MDS 3.0 assessment data, aggregated over 12 rolling months of resident observations. Short-stay measures mix MDS and Medicare fee-for-service claims data, depending on the specific measure. The claims-based measures (re-hospitalization, emergency department use) are processed through a separate CMS claims data warehouse pipeline on a longer lag — claims data can run 3–6 months behind real-time.
CMS applies risk adjustment to most quality measures using resident-level characteristics from MDS — age, gender, cognitive status, and diagnostic conditions. The adjustment is designed to make the rates comparable across facilities with different case mixes. Some measures are not risk-adjusted (e.g., antipsychotic medication use) because CMS considers the unadjusted rate to be the policy-relevant signal.
Quality measure rates are suppressed when the denominator (number of eligible residents) is below the minimum reporting threshold — typically 20 cases per year for annual measures, fewer for measures calculated over longer windows. A facility with fewer than 20 eligible residents for a measure displays 'data not available' on Care Compare for that measure, which can affect its quality measures star rating.
Ownership data: PECOS and cost reports
Ownership disclosure on Care Compare comes from two sources. PECOS (the CMS Provider Enrollment, Chain and Ownership System) captures ownership as reported at enrollment and at any Change of Ownership (CHOW) filing. PECOS requires all owners with 5% or greater interest to disclose their ownership type (individual, corporation, non-profit, government), percentage share, and relationship to any management company. This data is publicly accessible via the SNF All Owners dataset on data.cms.gov.
Annual cost reports (CMS-2540) require facilities to disclose related-party transactions, management company fees paid, and non-arm's-length arrangements. These disclosures are more granular than PECOS in some respects but are filed annually and are processed with a 6–12 month lag after the fiscal year closes. Fonteum's ownership networks study linked PECOS and cost report data to map chain affiliations, private equity ownership, and REIT relationships across 15,000+ facilities.
Neither PECOS nor Care Compare has historically required real-time disclosure of management company changes that fall short of a formal CHOW. This gap means the disclosed ownership on Care Compare can lag actual operational control by 12–18 months. The new CMS nursing home ownership transparency rule (effective 2024) tightens these disclosure timelines but does not yet require disclosure of private equity fund structures above the direct facility-owner level.
What gets withheld: suppression rules
CMS applies several suppression rules that limit what appears on Care Compare. Quality measures below the minimum denominator threshold appear as 'data not available.' Survey data under formal appeal to an Administrative Law Judge (ALJ) may be withheld from public display until the appeal is resolved — a process that can take 12–24 months. Facilities in the initial certification period (typically 12–18 months after first CCN issuance) receive a 'new to Care Compare' designation rather than a star rating.
Enforcement actions — civil money penalties, directed in-service training, state monitors — appear on Care Compare, but with limitations. Penalties under appeal are shown with a flag indicating the appeal status. Penalties for which escrow has been paid appear separately from penalties collected outright. Penalties imposed by states under their own Medicaid certification authority are not reflected on the federal Care Compare display.
The underlying federal datasets (available on data.cms.gov) have different suppression rules than the consumer display. Research downloads of the health deficiencies data include citation detail that is sometimes aggregated or paraphrased on the Care Compare consumer page. Fonteum accesses the underlying research files rather than the consumer display to ensure maximum data completeness.
Frequently asked questions
- How current is the data on CMS Care Compare?
- It depends on the data element. Health inspection data typically reflects the most recent standard survey (on average 11–12 months ago, since surveys occur annually). Staffing data is 3–4 months old at any point in the quarterly cycle. Quality measure data is calculated over a 12-month window and updated quarterly. Ownership data can be 12–18 months stale if no CHOW was filed. Fonteum's data freshness page shows the snapshot date for each dataset.
- Why doesn't Care Compare show all deficiency details?
- Care Compare shows deficiency text from the CMS-2567, but the display is designed for consumers, not researchers. Some citation details are summarized or formatted for readability. Records under ALJ appeal may be withheld or flagged. Researchers who need the full citation record should access the underlying CMS Health Deficiencies dataset on data.cms.gov.
- Is Care Compare data the same as what Fonteum uses?
- Fonteum ingests the underlying flat-file datasets published by CMS on data.cms.gov — the same data that populates Care Compare, but in research format with all fields present. Fonteum applies its own aggregations, chain-of-custody provenance tracking, and cross-dataset joins. The figures on Fonteum research pages are derived from the underlying federal files, not scraped from the Care Compare consumer interface.
- Why is private equity ownership not shown on Care Compare?
- Care Compare reflects PECOS-disclosed ownership, which requires disclosure of all parties with 5% or greater direct ownership interest. Private equity funds typically hold interests above the direct facility owner (through holding companies, management companies, or fund structures), and those upstream entities are not required to file in PECOS. The 2024 CMS ownership transparency rule expands disclosure requirements but does not yet mandate disclosure of fund-level PE structures.
Reviewed by Jennifer Montecillo, MD, medical reviewer. Non-practicing medical reviewer. Last reviewed: June 2026. Data last updated: 2026-06-03.
This explainer draws from federal primary sources. Every figure is traceable to a specific CMS dataset with provenance documentation. Data sources → Corrections log → Methodology library →
Federal data is in the public domain under U.S. Government Works. Fonteum analysis and synthesis are copyright Fonteum, Inc.