111 four- and five-star nursing homes had severity-G or worse deficiencies in the CMS three-year window

What this report is, and is not

The CMS five-star quality rating system, launched in 2008 and refined repeatedly since, is the federal government's primary tool for helping consumers compare nursing homes. Every facility receives one to five stars overall, driven by three sub-components: health inspection results, staffing levels, and quality measure scores. A higher star rating is a meaningful signal — and yet it does not map one-to-one onto the presence or absence of documented harm events.

This report documents that gap using CMS's own data. We joined the CMS Care Compare NH Health Deficiencies dataset — 418,148 deficiency citations across 14,635 nursing homes — with the CMS Nursing Home Compare star-rating snapshot. The join key is the CMS Certification Number (CCN). Every facility in the ranked table is identified by its CMS-published name and CCN, not a Fonteum assessment.

This report does not:

• Rate any facility independently. CMS conducts surveys; Fonteum reports the record.
• Imply that star ratings "caused" harm — they are a measurement tool with documented limitations.
• Recommend or warn against any specific facility for any specific resident.
• Attribute all deficiencies to inadequate care — surveyor variation, inspection timing, and facility size all affect the record.

What it does: it names the 111 four- and five-star facilities where CMS surveyors documented actual harm during the three-year window, ranks them by severity, and compares sub-rating patterns to help consumers and policymakers understand which star components carry the strongest signal about harm outcomes.

The CMS A–L severity scale: where "harm" begins

CMS classifies every deficiency citation on two axes: scope (isolated, pattern, or widespread) and severity (no harm, potential harm, actual harm, or immediate jeopardy). The intersection produces a letter code from A (isolated, no actual harm) to L (widespread, immediate jeopardy). For this study, "G+" means any citation at Severity G or above — the threshold where actual harm to a resident was documented.

Source: CMS State Operations Manual Appendix P — Survey Protocol, Regulations, and Interpretive Guidelines for Long Term Care Facilities.

G+ deficiency rate by overall star rating: the full distribution

Star ratings and G+ deficiency citations are correlated — lower-rated facilities have substantially higher rates — but no star tier is free of documented harm events. Among five-star facilities, 1.5% had at least one G+ citation. Among one-star facilities, the rate was 4.8%.

The gap between one-star (4.8%) and five-star (1.5%) is substantial — a four-fold difference. But the five-star rate being nonzero is the counterintuitive finding this study documents. Among the 3,032 five-star facilities in the dataset, 45 had at least one G+ citation — meaning a CMS surveyor found documented harm despite the facility earning the highest possible overall rating.

Which star sub-component is the weakest predictor of harm?

The CMS composite score is built from three components: health inspection, staffing, and quality measures. Each is independently rated one through five stars before being weighted into the overall score. To understand which component best distinguishes facilities with G+ deficiencies from those without, we compared average sub-ratings across three cohorts.

The health inspection sub-rating shows the largest drop — from 2.83 across all facilities to 2.10 for G+ facilities, a difference of 0.73 stars. This makes sense: the health inspection component is built from survey findings, and G+ deficiency citations are survey findings. They are not the same input — the inspection rating uses a more complex rolling formula — but they share the same upstream data.

The quality measures sub-rating shows the smallest drop — only -0.38 stars. QM ratings are computed from claims and minimum data set (MDS) assessments covering chronic-disease management, mobility, depression screening, and other resident outcomes. They do not directly measure survey deficiency findings. A facility with strong chronic-care management and poor infection control or medication administration practices can score high on QM while accumulating G+ survey citations.

Among four- and five-star facilities with G+ deficiencies specifically, the quality measures average is 4.54 stars — nearly the maximum. This cohort's high overall rating is driven substantially by QM performance and staffing inputs, while the health inspection sub-component averages 3.65 — still respectable, but lower. Consumers relying on star ratings as a harm filter should give disproportionate weight to the health inspection sub-rating, not the composite score.

Top 100 highest-rated facilities with severity-G or worse deficiencies

The table below lists the 100 four- and five-star nursing homes with the most G+ deficiency citations in the CMS dataset, sorted by G+ count descending. Where two facilities tie on G+ count, the higher overall rating ranks first. All facility names are CMS-published; identities resolved via the CMS Nursing Home Compare snapshot (2026-05-07). Columns: HI = health inspection sub-rating; ST = staffing sub-rating; QM = quality measures sub-rating.

Source: CMS Care Compare NH Health Deficiencies + CMS Nursing Home Compare snapshot (2026-05-07). G+ = severity G, H, I, J, K, or L. Worst = highest severity code in the three-year window. "Last G+" = most recent G+ survey date on record. CMS data; Fonteum does not independently inspect or rate any facility.

The inverse: 5,581 low-star facilities with zero documented harm deficiencies

The data-asserted gap runs in both directions. Just as some high-star facilities carry G+ deficiency records, 5,581 one- and two-star facilities — 95.4% of all low-rated homes — have zero G+ deficiency citations in the three-year window. These facilities' low ratings reflect deficiencies at the minimal-harm band (Severity A–F), staffing inputs below CMS benchmarks, or quality measure scores driven by chronic-care outcomes — not documented harm events.

This should not be read as a clearance of those facilities. A facility can accumulate many Severity D and E citations (potential harm, not actual harm) without crossing the G threshold. The absence of G+ citations means a CMS surveyor did not document actual harm over the three-year window — not that no harm occurred, and not that the facility's care standards are acceptable. Star ratings are a weighted composite; a low star rating with no G+ citations indicates a facility that is deficient in ways the composite captures but the harm threshold does not.

State distribution: where high-star + harm facilities concentrate

The distribution of four- and five-star facilities with G+ deficiencies tracks loosely with state size — larger states with more nursing homes produce larger absolute counts. The top four states (California, Illinois, Texas, and Ohio) account for 543 of the 111 facilities in this cohort.

State-level counts are not normalized for facility count or population. A state with 1,000 nursing homes will produce more absolute high-star + harm facilities than a state with 100, even if its rate is lower. Normalized rates (share of all facilities in the state that are 4-5★ AND have G+ deficiencies) are not shown here because they require state-specific total-facility counts, which vary by CMS snapshot vintage. The raw count is the figure that names real facilities in real places; the rate is the figure appropriate for state-to-state policy comparison.

Three-year trend: G+ citations peaked in 2024 and remain elevated

The CMS Care Compare deficiency dataset is a rolling three-year window — earlier surveys age out as new ones are added. Examining survey dates within the current window reveals that G+ citations are not diminishing.

G+ citations peaked in 2024 at 7,620 — a 29% increase over the 2023 figure. The 2025 figure of 6,636 represents a partial decline but remains elevated above 2023 levels. The 2026 figure (809 across 610 facilities through the snapshot date) represents only a fraction of the year. The trajectory does not indicate the star-vs-harm gap is closing. CMS survey resumption following pandemic-era suspension likely contributed to the 2024 spike.

For context: KFF research published in 2022 found that pandemic-year survey declines masked an accumulation of deficiencies that subsequently appeared in post-pandemic inspection cycles. GAO reporting in 2023 documented ongoing CMS staffing constraints that affect survey frequency and depth. This study's 2024 peak is consistent with both findings — resumed surveys exposing a backlog. It does not imply care quality deteriorated in 2024; it implies CMS was surveying more frequently and completely after a period of constrained activity.

Situating this study: what KFF, GAO, and LTCCC have published

The quality of CMS star ratings as a proxy for care quality is not a new question. Several peer-reviewed and policy research organizations have examined the issue from different angles; this study adds a dimension none of them provides: a named, ranked list of high-star facilities with G+ deficiency records.

KFF Health News (2022)

KFF Health News reporting documented that pandemic-era suspension of routine nursing home surveys created a backlog of uninspected facilities, some of which continued to receive high star ratings despite not being surveyed. The star-rating system uses a rolling inspection history; when inspections stopped, old scores decayed slowly. This created a period where ratings were based on aging data. Our study's 2024 citation spike is consistent with the resumption of surveys revealing deficiencies that had accumulated during that gap.

GAO (2023)

A 2023 GAO report on nursing home oversight found that CMS survey frequency and staffing constraints limited the government's ability to inspect facilities on a timely schedule. The report identified that a significant share of facilities went more than 12 months between standard surveys — longer than the CMS-required annual cycle. Facilities that are surveyed less frequently have more opportunity for undetected deficiencies to accumulate between inspections.

Long-Term Care Community Coalition (LTCCC)

LTCCC has published extensively on the limitations of the CMS star-rating methodology, arguing that the composite score obscures specific deficiency findings that are more predictive of resident safety. Their advocacy has focused on the health inspection sub-rating specifically, consistent with this study's finding that the HI component shows the largest drop among G+ facilities (−0.73 stars) relative to the population average.

What this study adds

None of the above sources publishes a named, facility-level list ranked by G+ deficiency count with sub-rating breakdown. The prior literature operates at the aggregate level. By joining the deficiency record to the star-rating file at the CCN level and surfacing the top-100 table, this study provides a reproducible, facility-level artifact that journalists, researchers, and consumers can use to investigate specific facilities using additional sources — CMS Care Compare, state survey reports, and the CMS complaint process.

Methodology and reproducible SQL

The analysis joins two CMS data artifacts at the CCN (CMS Certification Number) level:

1. CMS Care Compare NH Health Deficiencies — 418,148 deficiency citations across 14,635nursing homes (CMS source modified 2026-04-17, snapshot fetched 2026-05-08). Stored in Fonteum's nh_health_deficiencies table. The column scope_severity_code carries the CMS A–L severity code; G+ means codes G, H, I, J, K, or L.
2. CMS Nursing Home Compare star ratings — 14,699 facilities with overall_rating, health_inspection_rating, staffing_rating, and qm_rating per facility (snapshot 2026-05-07, stored on-disk).

Reproducible SQL for the per-CCN G+ counts:

SELECT
  ccn,
  COUNT(*) AS g_plus_count,
  MAX(survey_date) AS most_recent_date,
  MAX(scope_severity_code) AS worst_severity
FROM nh_health_deficiencies
WHERE scope_severity_code IN ('G','H','I','J','K','L')
GROUP BY ccn
ORDER BY g_plus_count DESC;

The high-star + harm cohort is constructed by joining the above result to the NH Compare snapshot on CCN, then filtering for overall_rating >= 4. The top-100 table ranks by g_plus_count DESC, overall_rating DESC.

Sub-rating averages are computed as simple means over non-null values. The “all facilities” denominator includes all 14,699facilities in the NH Compare snapshot. The “G+” denominator includes all 8,235distinct CCNs with at least one G+ citation matched to the NH Compare snapshot.

Year-by-year G+ counts use EXTRACT(year FROM survey_date) with a survey_date >= 2023-01-01 filter, consistent with the CMS rolling three-year window.

Limitations

• Snapshot timing mismatch.The NH Health Deficiencies snapshot was fetched 2026-05-08; the NH Compare star-rating snapshot is dated 2026-05-07. A facility's star rating as of the snapshot date may not reflect its rating at the time of any specific deficiency citation. Ratings update quarterly; deficiency citations affect the health inspection sub-rating on a rolling basis.
• Surveyor variation. The severity classification of a deficiency citation reflects the professional judgment of a CMS surveyor and the state agency conducting the survey. Severity-G thresholds may be applied differently across states and survey teams. High harm-rate states (see the NH Deficiency Harm Rate study) may reflect both care quality differences and survey intensity differences.
• G+ is a minimum, not a maximum. A facility with one G+ citation and a facility with 12 G+ citations are both in this cohort. The ranked table sorts by count to surface the most-cited facilities, but a single G+ citation is still a documented harm event.
• No causal claim. This study does not claim that high star ratings cause harm, prevent harm detection, or mislead consumers in any legally specific sense. It documents that two distinct CMS measurement systems — the star-rating composite and the deficiency severity record — produce discordant signals for a substantial share of facilities.
• Rolling window. The CMS deficiency dataset covers a rolling three-year window. Facilities at the boundary may have aged-out citations that are not captured. The 2026 partial-year figure is not comparable to full years.
• No Fonteum independent assessment. Every facility named in this report is identified from CMS-published data. Fonteum does not inspect, rate, verify, endorse, or guarantee any nursing home.

Frequently asked questions

Data sources

• CMS Care Compare NH Health Deficiencies — U.S. Government Work in the public domain. data.cms.gov/provider-data/topics/nursing-homes
• CMS Nursing Home Compare — Overall, health inspection, staffing, and quality measures star ratings. medicare.gov/care-compare
• CMS Five-Star Quality Rating System Technical Users' Guide (2024). Published by the Centers for Medicare & Medicaid Services.

Suggested citation: Fonteum Research Bureau. “111four- and five-star nursing homes had severity-G or worse deficiencies in the CMS three-year window.” Fonteum Research. Published 2026-06-04. https://fonteum.com/research/nursing-home-stars-vs-actual-harm