Methodology

Every grade on this site is computed from public CMS data by the formula below. No human adjusts an individual facility's score, and no facility can pay to change one. If you can obtain the same CMS files, you can reproduce every number we publish.

The base score

Each facility starts with a 0–100 composite of four CMS measures, weighted as follows:

ComponentWeightHow it is scored
Staffing compliance35% Reported RN hours per resident per day divided by 0.55, capped at 150% of the benchmark. A facility at or above 0.55 scores at least 0.67 on this component; one at 0.825 or higher scores the full 1.0.
Inspection clean rate30% 1.0 at a verified zero-deficiency current survey, falling linearly to 0.0 at 20 or more deficiencies. A missing inspection is never treated as zero.
Quality measures20% CMS quality star rating, normalized from 1–5 stars to 0.0–1.0.
Staffing consistency15% CMS staffing star rating, normalized from 1–5 stars to 0.0–1.0.

The weighted components are summed and multiplied by 100. That is the base score.

The 0.55 hour RN standard was repealed effective February 2, 2026 and is not currently enforced. We still grade against it. Why

Missing-data policy

Missing evidence is not a zero, and it is not a clean record. Every facility is assigned one of three evidence-completeness states before the grade is published.

StateMeaningGrade behavior
CompleteAll four Grade 1.x score inputs and current inspection evidence are present.Normal score and letter grade.
PartialCurrent inspection evidence is present, but RN staffing, the CMS quality rating, or the CMS staffing rating is unavailable.A conservative lower-bound score is shown. Missing components earn zero points and the remaining weights are not renormalized upward.
InsufficientThe current inspection/deficiency evidence required to distinguish a real clean survey from missing data is unavailable.The NursingHomeGrade score and letter are withheld and displayed as not rated.

This policy is intentionally asymmetric: missing evidence is never allowed to improve a score. A verified survey with zero deficiencies can earn the full inspection component; a missing survey cannot. Grade 2.0 will use the CMS Survey Summary source as an additional authoritative check because that file records inspection dates even when an inspection produced no citations.

Penalty terms

The base score is built from averages, which makes it blind to two facts that matter more to a family than any average: whether residents were actually harmed, and whether anything found is still unfixed. Two penalties are subtracted from the base score to correct that.

Uncorrected findings (up to 25 points)

Each deficiency the facility has not resolved costs:

1.5 × severity weight × plan weight × recency weight

The total is capped at 25 points.

Actual harm (up to 25 points)

Scored separately from raw deficiency counts. Each citation at scope and severity G through L costs 8 points for immediate jeopardy (J–L) or 4 points for actual harm (G–I), multiplied by the same recency weight. Capped at 25 points.

The no-plan rule

Independent of the arithmetic above: no facility holding any deficiency in "no plan of correction" status can be graded A. Such a facility is capped at B regardless of its score. This rule only ever lowers a grade — it never raises one.

Grade bands

GradeScore
A80–100
B65–79
C50–64
D35–49
F0–34
NRNot rated because critical current inspection evidence is insufficient

Update cadence and freshness

CMS publishes the nursing-home datasets we use on a recurring schedule, while individual health inspections occur on a rolling basis. We re-ingest the CMS files and recompute all grades when a new source release is loaded.

We track the CMS source-modified date, CMS public release date, NursingHomeGrade import timestamp, and CMS processing/data-period date separately. See data sources for the release and import dates currently in production.

Sitemap freshness uses the CMS source-modified date rather than the import clock. Rerunning an unchanged ingest therefore does not manufacture a newer lastmod date.

Grade 2.0 research track

Grade 2.0 is being built and validated separately from the public Grade 1.x formula. Its Phase A evidence layer adds CMS Survey Summary, measure-level MDS quality measures, Medicare claims quality measures, and richer case-mix-adjusted staffing data. Those sources are stored in shadow tables and do not change the public grade until a versioned model passes temporal validation and migration review.

Grade changelog

DateChange
2026-09-05 Added explicit complete/partial/insufficient grading semantics. Missing inspection evidence now withholds the grade instead of receiving clean-inspection credit; other missing components produce a labelled lower-bound partial score without weight renormalization.
2026-07-27 Added the uncorrected-findings and actual-harm penalty terms and the no-plan rule. Of 14,703 facilities, 3,858 (26%) moved down at least one letter grade and 10,845 were unchanged. The number of A grades fell from 2,677 to 2,184. No facility's grade improved as a result of this change.

If you believe a grade is wrong

We publish a corrections policy and a dispute route. Grades are derived from CMS data, so a grade we compute incorrectly and a CMS record that is itself wrong are different problems with different fixes — the contact page explains both.

Related

The plain-language version, the underlying files, and the staffing benchmark.

How we grade → Data sources → The repealed staffing rule →