A federal nursing home staffing floor was repealed before it was ever enforced — CMS stated in the 2024 rule that more than 79 percent of facilities would have to increase staffing to meet it, and Public Law 119-21 bars enforcement through 2034-09-30

What is happening?

The United States has no numeric federal staffing floor for nursing homes. CMS finalized one on 2024-05-10 — a registered nurse onsite 24 hours a day, 7 days a week, plus 3.48 total nurse hours per resident day, of which 0.55 had to come from registered nurses and 2.45 from nurse aides — and repealed it on 2025-12-03, effective 2026-02-02, before a single hours floor had ever bound a facility. Measured against the CMS Care Compare staffing file processed 2026-07-01, 1,023,198 residents per day — 82.7 percent of residents in facilities reporting all three measures — live in a nursing home that misses at least one of the three repealed floors.

Three instruments removed it in nine months, and each one could point at another. Two federal district courts vacated the staffing provisions at summary judgment in 2025, the court in the Northern District of Texas resting on the major questions doctrine. Section 71111 of Public Law 119-21, enacted 2025-07-04, then barred the Secretary of Health and Human Services from implementing, administering or enforcing the 2024 amendments to 42 CFR 483.5 and 483.35 through 2034-09-30. CMS deleted the text from the Code of Federal Regulations by interim final rule issued without prior notice and comment, on a good cause finding that keeping unenforceable provisions on the books would be confusing and impracticable.

The duty survives; the number does not. As of 2026-08-08 the operative text of 42 CFR 483.35 requires a registered nurse for at least 8 consecutive hours a day, 7 days a week, and still opens by requiring sufficient nursing staff to assure resident safety, measured against a facility assessment the facility writes for itself. A surveyor can still cite a facility for insufficient staffing. Nothing now states what sufficient is.

Whose problem is this?

The body that would have to restore a number is the same body that wrote one and then deleted it, and it is barred by statute from enforcing the version it wrote until 2034. Nothing outside that arrangement obliges anyone to produce a different one.

Where does this problem end?

The boundary matters here because the standard was written, finalized and then removed before it ever bound anyone. What is missing is not evidence and not a proposal but a number that a surveyor can measure a facility against.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here names a staffing level that would count as sufficient, a date by which any level is to be reached, or a metric anyone is accountable for. The 2001 CMS report to Congress cited by the state Attorneys General identified 4.1 hours per resident day as the level below which staffing-induced quality problems appeared, but no document opened in this round adopts that figure or any other as a target for the period after 2026-02-02.

How big is it?

1,023,198 residents per day. That is the sum of reported average daily census across the facilities that miss at least one of the three repealed floors, computed for this document from the CMS Care Compare provider file processed 2026-07-01. It is 82.7 percent of the 1,236,706 residents in facilities that report all three staffing measures.

The figure moves with the question asked. Read narrowly as residents in facilities below the single aggregate floor of 3.48 total nurse hours per resident day, it is 483,925 per day, or 39.1 percent. Read broadly as residents who lost an enforceable numeric floor regardless of where their facility sits today, it is 1,258,267 per day — every resident in every certified nursing home, including those in a facility at 3.60 hours per resident day that now has no obligation to stay there.

The spread is not uncertainty in the data. The same file produces all three counts. It is a choice about what the repeal is taken to have removed.

Two limits sit on the count itself. It is a daily census rather than a count of persons, and nursing homes admit and discharge continuously, so the number of distinct people passing through in a year is substantially larger and is published by no source opened here. And severity is flattened: a facility 0.05 hours below the floor counts the same as one 2.0 hours below, while the median shortfall among the 5,139 facilities under 3.48 is 0.25 hours per resident day.

Under what conditions does it arise?

1. A qualitative duty cannot be enforced against a facility that disputes it. The word sufficient is adjudicated case by case against a facility assessment the facility writes for itself under 42 CFR 483.71. A numeric floor moves that dispute from judgment to arithmetic. The Long Term Care Community Coalition reports that about 95 percent of substantiated nursing home health violations are classified as causing no harm, and that enforcement action follows in about half of the cases CMS identifies as actual harm or immediate jeopardy.

2. The standard was attacked at three independent points and only had to fail once. Courts vacated it, Congress suspended it, and the agency then deleted it. At no step in that sequence was the question on the table whether more than a million residents a day in facilities below the floor was acceptable.

3. The moratorium outlasts any current majority. Section 71111 bars enforcement until 2034-09-30, a little over nine years from enactment. A one-year or two-year pause would have left the rule standing in the Code of Federal Regulations and the argument alive. Nine years is what made deleting the text sound like housekeeping, and the housekeeping argument is what carried the good cause finding that let CMS skip notice and comment.

4. The cost accounting has no field for the loss. The repeal books USD 5,412 million a year in benefits at a 3 percent discount rate, USD 321 million a year in costs, and zero dollars in transfers. Nursing wages that will not be paid are recorded as a benefit rather than as a transfer from workers to facilities, and no health outcome in either direction is monetized or listed as unquantified.

5. The workforce argument and the compliance data point in different directions. CMS cited rural and tribal closure risk and a projected national shortage of 295,800 nurses. On the metric the rule actually used, rural facilities were slightly more likely to meet all three floors than urban ones as of 2026-07-01 — 24.3 percent against 20.8 percent — though rural facilities are smaller and carry lower census, which mechanically lifts hours per resident day, so this does not by itself answer a cost or recruitment argument.

What has been tried?

Two directions were tried at once — write a number, and litigate it away. The second finished first, and a third, the statutory moratorium, arrived before the earliest hours floor was due to bind. Six months after the comment period on the repeal closed, no further document has been published under the same regulation identifier.

What was found?

Why is it still unsolved?

Enforcement absent — the duty was never repealed, only the version of it that a surveyor could measure.

Sections 1819(b)(4)(C)(i) and 1919(b)(4)(C)(i) of the Social Security Act still require sufficient nursing staff, and 42 CFR 483.35 still opens with the same words. What was removed is the only form of that duty that does not require an argument first. Under the qualitative standard, whether a facility is adequately staffed is settled case by case against an assessment the facility itself wrote, which is why removing the number costs nothing on paper and why the industry was willing to spend years in court to stop it. The verbs Congress chose are implement, administer and enforce.

The second part is that the removal had three authors and no owner. The courts held that the agency had exceeded its statutory authority. Congress held that the rule should not operate for nine years. The agency held that keeping unenforceable text in the Code of Federal Regulations was confusing. Each of those is a defensible statement about an instrument, and none of them is a statement about staffing. Nobody in the sequence had to defend the condition that the number described.

The third part, which is what keeps the pattern stable, is that deleting the text raised the cost of coming back. Section 71111 is instrument specific — it bars enforcing the amendments made by one dated rule to two named sections, which is why 18 state Attorneys General could ask CMS for a replacement regulation without asking Congress for anything. But the old text is gone rather than dormant, so a replacement now needs a fresh record, a fresh regulatory impact analysis and fresh exposure to the same major questions holding that vacated the first one. The repeal did not merely comply with the moratorium. It made the return trip longer than the moratorium itself.

What observation would mean it is solved?

Candidates — (a) a federal numeric staffing floor is in force and is being cited by surveyors (b) the share of residents in facilities below 3.48 total nurse hours per resident day falls for several consecutive years (c) enforcement action follows a larger share of substantiated deficiencies.

(a) alone is weaker than it looks. Section 71111 bars enforcement of one dated rule until 2034-09-30, so a floor could exist on paper in a differently drafted rule and still spend years in litigation under the same major questions holding that vacated the first one. A rule in force is also not a rule applied — the 2024 rule stood in the Code of Federal Regulations from 2024-05-10 until the repeal took effect on 2026-02-02, and CMS states in the repeal preamble that its provisions had not been enforced.

(b) alone measures the wrong thing. Hours per resident day is a ratio, and it rises when residents leave. A facility that loses census faster than it loses staff improves on this metric while delivering less care. The measure also flattens severity, and the median facility below the floor as of 2026-07-01 was 0.25 hours short.

(c) alone counts process. More enforcement actions against a qualitative duty still leaves each one to be argued facility by facility, and the reported figures suggest the binding constraint is classification rather than volume — if about 95 percent of substantiated violations are recorded as causing no harm, the penalty question is settled before it is asked. The three have to be read together, and (b) has to be read against the census it came from.

What is it connected to?

Fills with researchstate staffing floors and how many residents they cover, Medicaid nursing home payment rates, the nurse and nurse aide labor market, substitution toward home and community based care, and the major questions doctrine as it applies to other health rulemakings. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• How many residents are in facilities below the floors. The repeal analysis counts facilities exactly once, at 14,752, and only to price the cost of reading the rule at USD 48.56 each. It never crosses facility compliance with resident census. The 2024 rule cites 1.2 million residents a day and does not cross it either. The figure of 1,023,198 in this document exists because it was computed here; no source opened in this round publishes it.
• What the standard was worth. The repeal accounting statement records USD 5,412 million a year of benefits, USD 321 million a year of costs and zero dollars in transfers. No health outcome appears in either column, not even as an unquantified item. The corresponding table in the 2024 rule is published in the Federal Register as an image, so the two ledgers cannot be compared without optical character recognition, and that comparison was not made here.
• Whether a replacement rule is lawful. This is the most consequential open question in the record and no source answers it. Section 71111 is instrument specific by its terms; 18 state Attorneys General urge a replacement regulation, which presupposes one is permitted; CMS says only that the repeal allows for future rulemaking and commits to nothing. No agency interpretation and no court decision opened here resolves it.
• A number in the justification that cannot mean what it says. The repeal preamble states that federal workforce analysts project just 63,720 people working as full-time registered nurses in 2030. The United States has on the order of three million registered nurses, so the figure is almost certainly a projected shortfall rather than a headcount. It sits in the section justifying the repeal, it is wrong in the direction that supports the action, and no correction and no commenter objection appears in anything opened here.
• Whether CMS will answer the comments. The comment period closed 2026-02-02. As of 2026-08-08 the Federal Register lists four documents under the regulation identifier 0938-AV25 and the newest is the repeal itself. An interim final rule normally ends in a final rule responding to comments, and no source opened here states a schedule for one or says whether one is planned.
• What happens on 2034-10-01. The moratorium expires that day and the regulation it suspended no longer exists to return. No statute, agency document or comment letter opened here says what, if anything, is scheduled to occur. The end date is stated everywhere and its consequence nowhere.
• How many workers this is. The entire monetized benefit of the repeal is USD 5.51 billion a year of nursing labor that will not be purchased. No document opened here converts that money into hours, positions or people, and the phrase full-time equivalent does not appear in the 2024 final rule text.
• How much of the country a state floor covers. The National Consumer Voice guide to its own state chart records that states count direct care incompatibly, several including therapy staff and a few including psychiatric coordinators, social services directors, infection preventionists and physicians. The organization best placed to produce a national coverage figure explains why it would not mean anything, and no source opened here produces one.

See the evidence

No agency press office was reachable, and both federal primary documents were read in full. The statutory text, the repeal preamble, the 2024 final rule and the current text of 42 CFR 483.35 were read directly rather than through any summary, and the moratorium and its 2034-09-30 end date appear identically in the statute, in the repeal preamble and in the GAO report. The compliance and exposure figures — 21.8 percent, 82.7 percent, 1,023,198 — are a computation performed for this document over the CMS provider file and are published in that form by no source; they are reproducible rather than cited, and are graded medium above for that reason. Where sources disagree the disagreement is left visible rather than resolved. The ten-year price of the same regulation appears as a cost of USD 43.0 billion in 2021 dollars and as a saving of USD 55.1 billion in 2024 dollars over a different window, and neither document reconciles the two. The share of facilities meeting all three floors is 21.8 percent on reported staffing columns and 26.2 percent on case-mix adjusted columns, and no source states which series compliance would have been judged against. The figure of about 90 percent failing, from the Long Term Care Community Coalition, is a different benchmark rather than a different fact, since it measures staffing expected from resident acuity rather than the three flat floors. Two federal press offices returned HTTP 403 to every attempt, so the departmental framing of the repeal is represented here only at second hand. This is a Path A output (research-based definition), so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 483,925–1,258,267 As of 2026-07-01

Sensitivity The width from 483925 to 1258267 is not uncertainty in the data, because the same file produces all three counts. It is a choice about what the repeal is taken to have removed: the low bound counts residents where total nurse hours are short, the central figure of 1023198 counts residents in facilities the rule would have obliged to change, and the high bound counts every resident who lost an enforceable numeric floor. What this number fails to count: harm, because the standard was never enforced and there is no before and after to measure; workers, whose unpurchased hours are the entire monetized benefit of the repeal and are converted into hours, positions or people by no document opened here; distinct persons rather than daily census, a figure no source opened here publishes; and severity, since a facility 0.05 hours below the floor counts the same as one 2.0 hours below, while the median shortfall among the 5139 facilities under 3.48 is 0.25 hours per resident day. The limit in the opposite direction: reported hours per resident day is a quarterly average, and the repeal preamble states that its own model assumed facilities would reallocate staff to stay compliant continually, so a facility compliant on a quarterly mean can be short on many individual days. Compliance measured continuously would therefore sit at or below the 21.8 percent found here and the exposed population at or above 1023198. Switching from reported to case-mix adjusted staffing columns moves compliance to 26.2 percent and the exposed population to 973040, a swing of about 5 percent, and no source states which series compliance would have been judged against.

Regional breakdown State and regional values were not confirmed in this round. The provider file carries a state field, but no source opened here publishes a state-level count of residents in facilities below the repealed floors, and the only alternative frame, state staffing floors, cannot be aggregated because the National Consumer Voice records that states define direct care incompatibly. Splitting the national figure by population share is not permitted and was not done.