What SeniorCareScore is

SeniorCareScore is an independent 0–100 score that summarizes publicly available quality, staffing, and regulatory data for U.S. skilled nursing facilities (SNFs). Our goal is to make it easier for families to compare nursing homes using the same public data CMS already publishes — organized into a single number, with the full component breakdown shown alongside it so you can see exactly why a facility scored the way it did.

SeniorCareScore is not a CMS rating. CMS publishes its own 1–5 star rating. Our score uses CMS data as an input but combines it with citation severity, staffing turnover, infection-control history, and recency of inspection in a way CMS’s star rating does not. A facility can have a high CMS star rating and a lower SeniorCareScore — this is by design, and the breakdown will show you which components pulled the score down.

How the score is built

SeniorCareScore v1.3 blends 8 components. Each component is scored on its own scale, then combined using the weights below. The final score is clipped to the 0–100 range.

Component weights (v1.3)

Total positive weight: 100. IJ penalty applies on top as a deduction of up to 20 points — a facility with recent Immediate Jeopardy findings cannot “average out” that signal with good scores elsewhere.

What changed from v1.1 to v1.2

No weights changed. v1.2 recalibrates three components — turnover, fines, and infection control — from fixed magic-number cutoffs to self-calibrating national percentile benchmarks, the same pattern citations have used since v1.1.1. A facility’s component score reflects how it ranks against the rest of the country today rather than against thresholds we wrote down a year ago.

How citation severity is grouped

CMS inspectors assign every citation a “scope and severity” code. SeniorCareScore groups those codes into three plain-English tiers so consumers can read a facility’s history without learning a new alphabet: Minor (paperwork or process issues, no resident harm), Serious (a resident was harmed), and Critical (residents were in immediate danger of serious injury or death — CMS calls this tier “immediate jeopardy”).

CMS’s underlying A–L letter codes remain available for journalists and researchers at the CMS Scope and Severity Grid.

Score labels

Labels are a convenience. The underlying number and the component breakdown are what matter — two facilities with the same label can have very different profiles.

Confidence gating

Not every facility has enough public data to score reliably. v1.2 uses two hard gates: a 30-bed minimum and 24 months of CMS-reported history. Facilities that do not meet both thresholds are flagged Low confidence and excluded from default leaderboard results.

If a facility is missing a required data element entirely (e.g., no reported staffing), the component defaults to a neutral mid-range value rather than a zero. This prevents a missing-data gap from being misread as a quality problem.

Data sources

All inputs to SeniorCareScore are public federal records.

Data freshness

Every API response includes a cms_data_as_of field. This is the date of the CMS publication we ingested — not the date we ran our scoring job. Always read this field before treating a score as current.

Known limitations

1. Public data lags reality. A citation that just happened won’t appear in CMS data until the survey report is finalized and published.
2. Self-reported fields are self-reported. Staffing and turnover come from facilities, verified through CMS’s PBJ payroll-based journal system but still facility-originated.
3. Citations reflect what inspectors find. A facility inspected infrequently may look cleaner than one that receives a thorough survey. Survey recency is included as a 5% component partly to surface this.
4. Small sample sizes are noisy. Even with the 30-bed and 24-month gates, rural or specialty facilities may have citation counts low enough that a single inspection swings the score.
5. We do not model resident acuity. A high-acuity sub-acute rehab may accumulate more clinical citations independent of care quality.
6. We do not model chain ownership risk beyond what’s in CMS data.
7. State variation. State survey agencies differ in how aggressively they cite.

Update and versioning policy

Minor refreshes (data only). Re-ingesting CMS data updates scores; methodology version stays the same. The cms_data_as_of field moves forward.

Minor methodology updates (v1.1 to v1.2). Weight tweaks, threshold changes, or new data-quality gates. New row in methodology_versions; prior version preserved.

Major methodology updates (v1 to v2). Structural changes — new components, removed components, changed definitions. Prior version runs in parallel for at least 90 days.

Corrections process

If you believe the underlying CMS data is wrong

The correct path is to work with CMS directly through the survey agency in your state — we do not override CMS data. Once CMS publishes a correction, it appears in our next refresh.

If you believe our calculation is wrong

If you are a facility operator, family member, or researcher and you believe a SeniorCareScore is wrong because of a calculation error, miscoded component, or bug, please contact us.

Email: corrections@seniorcarescore.com
Include: the CMS Certification Number (CCN), the URL of the page in question, and a description of the issue.

We commit to:

• An acknowledgment within 2 business days.
• A first-pass investigation within 10 business days.
• If we confirm an error, a public correction note in the facility’s score history.

Frequently asked questions

OSHA workplace-safety signal (v1.3)

v1.3 adds federal OSHA workplace-safety enforcement as a flag-only signal. The composite 0–100 score and its 8 component weights are unchanged from v1.2 — OSHA data surfaces directly on the facility page as flag pills, not as a hidden weight buried inside the score.

What we surface

• OSHA willful or repeat violations (last 5 years). Shown as a critical flag. Federal regulators use these classifications when they conclude an employer intentionally violated the standard or was previously cited for a substantially similar violation.
• OSHA penalties at or above $10,000 (last 5 years). Shown as a serious flag with the dollar total.

How facilities are matched

OSHA enforcement data does not contain CMS Certification Numbers. We match each OSHA inspection to a CMS-certified SNF using a five-tier deterministic matcher (tiers A-C are exact-match; tiers D-E use trigram similarity against the CMS legal business name and parent chain to catch operator-LLC filings that exact match misses):

1. Tier A — address + ZIP + state exact (after normalization). Highest confidence.
2. Tier B — address + city + state.
3. Tier C — name + city + state. Jaccard token similarity at or above 0.80, requires a shared core token, conservative threshold to avoid cross-facility false positives.
4. Tier D — multi-source name + address within state and 5-digit ZIP. Trigram similarity computed against the maximum of provider name, CMS legal business name, and parent chain name; combined with address similarity (weighted 0.55/0.45). Three sub-tiers with different thresholds: D1 (combined ≥ 0.80), D2 (0.75 to 0.80 with strong name or address anchor), D3 (address similarity ≥ 0.90 catches DBA/rebrand at exact site).
5. Tier E — name ≥ 0.85 with exact street number match within state and 5-digit ZIP. Catches the same building when address formatting differs materially (e.g., "1ST" vs "FIRST", "ROAD" vs "HIGHWAY").

The matcher is filtered to nursing-care NAICS code 623110 plus legacy SIC codes 8051, 8052, and 8059 to keep the universe to skilled nursing and intermediate care facilities. Inspections at non-care sites that happen to share an address (e.g., medical office buildings) are excluded.

Current match rate: 48.6% of CMS-certified SNFs link to at least one OSHA record. The remainder of CMS facilities have no federal OSHA inspection record at all (state-plan states like California and Oregon report only partially in the federal export). A facility with no flags is not affirmatively safe — absence of flags can mean a clean record, no recent OSHA inspection, or a record we did not match with high enough confidence.

Limits

• OSHA covers worker safety, not resident care. The flag is a workplace-safety signal that often (but not always) correlates with operational discipline.
• Many states run their own state-plan OSHA programs. State-plan inspections are included in the federal feed used here, but the dataset reflects the federal aggregation lag.
• Penalties shown are initial assessments. Negotiated reductions are reflected when OSHA publishes the updated record.

Changelog

• v1.3.1 — May 3, 2026. OSHA matcher expanded from 3 to 5 tiers using CMS legal business name and parent chain. Match rate improved from 47.7% to 48.6%; +141 facilities now have OSHA flags. No score-weight changes; flag presentation unchanged.
• v1.3 — May 3, 2026. Adds OSHA workplace-safety flag signal (willful/repeat violations and 5-year penalty totals from federal OSHA enforcement records). Three-tier deterministic facility match. No score-weight changes from v1.2.
• v1.2 — May 1, 2026. Recalibrated turnover, fines, and infection-control components from fixed magic-number cutoffs to self-calibrating national percentile benchmarks. No weight changes. Added explicit legal-defense block, expanded data-source URLs, formal corrections workflow, and FAQ.
• v1.1 — April 22, 2026. Initial public methodology. 8 components, 30-bed and 24-month confidence gates, IJ penalty up to −20.

SeniorCareScore is an independent informational tool operated by Blue Heron Works, Inc. Underlying data © U.S. Centers for Medicare & Medicaid Services. Methodology © Blue Heron Works, Inc. This document is not legal or medical advice.