Loading…
Understanding our harm-related evidence composite and state peer percentile
The SunsetWell Score is a 0-100 descriptive composite that combines current and historical signals in public CMS data for Medicare-certified nursing homes. It does not estimate a resident's probability of future harm. We provide two separate metrics:
Harm-related evidence composite
How this facility compares to peers in your state
What it summarizes: How favorable is the combined picture from inspection severity, staffing, CMS clinical measures, and complaint and incident data?
Example: A score of 78 falls in the "Excellent" band because the facility's combined CMS evidence is favorable under the v2.3 weighting method.
Key point: This score is not a state percentile. It is calculated the same way in California and Wyoming. It is also not a probability: 78 does not mean 78% safe or 78% less likely to have an adverse event.
What it measures: How does this facility compare to other nursing homes you might be considering in the same state?
Example: 85th percentile in California means this facility ranks better than 85% of California nursing homes (top 15%).
Facility pages also show an Externally Measured Outcomes score (0-100). It is built from CMS's claims-based quality measures — hospital admissions and emergency-department visits per 1,000 long-stay resident days, short-stay rehospitalizations and ED visits, potentially preventable readmissions, discharge back to the community, and infections requiring hospitalization. These come from Medicare billing records, not from assessments the facility submits about itself, which makes them a useful outside check on self-reported data.
Because small facilities have few residents, we shrink each measure toward the national median in proportion to its sample size before ranking, so a handful of unusual months has less effect. Facilities need at least two claims measures to receive a score at all; otherwise we say "insufficient data" rather than guess.
The "Two ways of measuring this home" comparison on facility pages puts the self-reported view and the Medicare-records view on the same 0-100 scale. When they diverge by 40 points or more in either direction, we flag it — worded as a question to ask, not an accusation, because timing differences and resident mix can explain part of any gap. In our June 2026 data, roughly 3% of nursing homes diverge in each direction.
The claims-based score is intentionally kept separate from the SunsetWell Score above: the two draw on different data channels, and our own analysis shows they are largely independent signals — which is exactly why showing both tells you more than blending them would.
The SunsetWell Score and Percentile Rank answer different questions:
SunsetWell Score answers:
"What does the combined public evidence show under one consistent method?"
Percentile Rank answers:
"How does this facility compare to my other options in this state?"
Facility in San Francisco, CA
The evidence composite is favorable, while many California facilities have similar or higher results. Compare the underlying measures before deciding.
Facility in rural Wyoming
The same composite result ranks more highly among Wyoming peers. The percentile describes local context, not whether the facility fits an individual resident's needs.
We analyze quality metrics from:
All data comes from publicly available federal sources.
We use a reproducible statistical weighting method to combine public CMS data for approximately 14,700 Medicare-certified nursing homes. The result is a descriptive composite of several harm-related evidence channels, not a forecast of what will happen to an individual resident.
What we include: The composite uses inspection severity, staffing depth and stability, CMS clinical measures such as falls with injury and pressure injuries, and facility-level complaint and incident rates.
The v2.3 weights were fitted against a same-period composite harm index made from clinical measures, inspection severity, and federal penalties. That design makes the score a harm-weighted summary; it does not establish that the score forecasts future outcomes better than simpler alternatives.
Our approach: We normalize the inputs, address missing values using documented rules, apply the published v2.3 weights, and map the result to a 0-100 scale. Because complaint rates are an input, same-period complaint correlations are a consistency check rather than independent validation.
Official Federal Rating
CMS publishes official overall and domain ratings for Medicare- and Medicaid-certified nursing homes using health inspections, staffing, and quality measures under its published methodology.
When to use it:
Use it as the official federal baseline and review the inspection details behind the stars
Descriptive Evidence Composite (v2.3)
A 0-100 research composite built from inspection, staffing, clinical-measure, complaint, and incident data, with a separate state peer percentile. It is not an official federal rating.
When to use it:
Use it to compare the combined evidence and then inspect the underlying measures
No. CMS 5-star ratings are the only official federal ratings. Our score is analytical research based on publicly available CMS data.
CMS produces the official 1-5 star rating from health inspections, staffing, and quality measures. SunsetWell uses a different weighting method and also includes complaint and incident rates, so the results can disagree. A disagreement is a reason to review the component data and current inspection details, not proof that either rating captures the whole picture.
The June 2026 same-period comparison found an association between the SunsetWell Score and size-adjusted substantiated complaints (Spearman ρ = 0.41), below CMS Overall stars (0.46) and Health Inspection stars (0.48), and above CMS Staffing (0.25) and Quality Measure stars (0.09). Because complaints are also an input to the SunsetWell Score, this is a consistency check, not independent validation. In July 2026, a temporal evaluation tested the 10 of 25 v2.3 inputs that could be reconstructed from October 2025 data against the following seven months. That subset had a modest association with later citations, but it did not outperform the facility's own past deficiency severity. Staffing was not available for that test, and the model parameters had been fit using newer data, which favors the composite. The evaluation was a v2.4 research gate; no v2.4 score was released and no score changes resulted. We therefore describe the public v2.3 score as a descriptive composite, not a forecast.
Facilities cannot pay SunsetWell to change their score or position. The method uses published CMS inputs and documented weights. As with any system that relies partly on reported measures, the score is not immune to source-data limitations; review inspections, staffing, claims-based outcomes, and facility-reported measures together.
Neither rating should make the decision by itself. CMS stars are the official federal rating; SunsetWell is an independent research composite that organizes some of the same evidence differently. Read the underlying measures, ask what has changed since the data date, and confirm that the facility can meet the resident's needs.
We recalculate scores with each CMS data refresh. Our facility data was last refreshed in June 2026.
Our v2.3 model achieves complete nursing-home coverage through robust handling of missing data.
Assisted living data is different: most states publish far less comparable inspection and outcome data for assisted living than CMS does for nursing homes. Assisted living listings therefore do not receive a comparable numeric SunsetWell quality score.
Scores are recalculated with each CMS data refresh (data last refreshed June 2026).
Current public score version: v2.3 descriptive composite (launched October 2025). The July 2026 v2.4 work was an evaluation only; it did not release or change a public score.
Not a forecast or guarantee: The score summarizes public data; it does not estimate what will happen to a particular resident. Care quality depends on many factors that available data cannot measure.
Visit in person: Use our score as a starting point, but always visit facilities in person, talk to staff, and trust your instincts.
Individual needs vary: A high-scoring facility might not be the right fit for your specific situation. Consider your loved one's medical needs, location preferences, and family priorities.
SunsetWell Scores are analytical tools based on publicly available data and should not be considered medical advice. Always consult with healthcare professionals and visit facilities in person before making placement decisions. Past performance does not guarantee future results.
CMS 5-Star Ratings are the official federal quality ratings for nursing homes. For official quality information, visit Medicare.gov.
Search 73,000+ care listings. Medicare-certified nursing homes include SunsetWell Scores and public evidence where available.
Start Your Search