The agency-staffing line: how much of US nursing-home nurse care now comes from temp staff

Every figure in this study comes from one frozen federal source: the CMS Payroll-Based Journal, the daily staffing record that every Medicare- and Medicaid-certified nursing home in the country must file from its own payroll system. We read the 2025 Q2 snapshot — the quarter running April 1 through June 30, 2025 — and asked a single question the file is uniquely able to answer. Of all the nurse hours worked in America's nursing homes that spring, how many were bought from a staffing agency rather than worked by the home's own employees?

The answer is 5.98%. Out of 423.7 million nurse-staffing hours logged across 1,322,254 facility-days with residents present, 25.3 million were supplied by contract or agency staff. That is roughly one nurse hour in every seventeen, bought on the temporary market. It is not a survey estimate or a model. It is a sum of columns in the payroll file.

The short answer. In the CMS Payroll-Based Journal's 2025 Q2 snapshot, 5.98% of US nursing-home nurse-staffing hours — 25.3 million of 423.7 million — came from contract or agency staff rather than direct employees. Reliance ranged from 25.1% in Vermont to 1.0% in Alabama, a roughly 25-fold state spread. Eighty-six individual homes bought more than half their nurse hours from agencies.

The reason this number matters now is regulatory. The federal minimum-staffing standard for long-term-care facilities — finalized in 2024 — was rescinded, with the rescission effective February 2, 2026. For the first time in the modern history of the program, there is no national floor under nurse staffing. Agency reliance is the pressure valve facilities reach for when they cannot fill permanent lines, and the figures below are the pre-rescission baseline against which the next several quarters will be read.

What the contract column actually records

The Payroll-Based Journal, or PBJ, is the most granular public record of who worked in a nursing home and when. Each row is one facility, one calendar day, with hours broken out by job category: registered nurses, licensed practical nurses, certified nursing assistants, nurse aides in training, and medication aides, along with the day's resident census.

What makes this particular question answerable is a detail of how PBJ is structured. For every nurse job code, CMS requires the facility to report two paired numbers — hours worked by its own employees, and hours worked by contract or agency staff. For registered-nurse direct care, those columns are Hrs_RN_emp and Hrs_RN_ctr. The same employee-versus-contract split exists for all eight nurse job codes the file tracks. Our agency share throughout this report is the sum of the eight contract columns divided by the sum of all nurse hours, restricted to facility-days where residents were actually present.

The word "contract" here means staff sourced through a third-party staffing agency and billed to the facility, rather than carried on its own payroll. It is the nursing-home equivalent of a travel nurse or a per-diem agency shift in a hospital. The distinction is not a bookkeeping nicety. An agency nurse is paid more per hour, has no long-term relationship with the residents, and may never have worked that building before. Facilities turn to agency staff when they cannot recruit or retain permanent nurses — which is why the share is read, across the staffing literature, as a barometer of workforce instability rather than as a neutral staffing choice.

How big the agency line is, nationally

At 5.98% of all nurse hours, the agency line looks modest in aggregate. It is larger than it sounds once set against the size of the system: 25.3 million hours is the equivalent of tens of thousands of full-time nurses' worth of work, last spring, that no nursing home actually employed. Translated to the standard staffing measure, agency staff delivered about 0.23 of the 3.78 total nurse hours per resident-day the file records.

The share is not uniform across roles. Licensed practical nurses — the workhorse medication-and-treatment role in most homes — were the most agency-dependent, at 7.37% of LPN hours. Registered nurses came next at 5.78%, and the aide categories that deliver the bulk of hands-on personal care were lowest at 5.51%. That ordering is itself telling: the licensed roles, which are harder to recruit and command the steepest agency premiums, are exactly where homes most often reach for temporary coverage.

The geography of the agency line

Sort the same hours by state and the national average dissolves into a map of enormous disparity. The table below shows the nine states with the highest agency share of nurse hours, and the five lowest, among states reporting in the quarter.

In Vermont, a quarter of every nurse hour worked in a nursing home was bought from an agency. In Alabama, it was one hour in a hundred. That roughly 25-fold gap is far wider than differences in resident need can explain on their own. It points at the things states actually shape — Medicaid reimbursement rates, the depth of the local nurse labor market, the presence or absence of state staffing rules layered on top of the federal floor, and how rural the geography is. The PBJ file cannot adjudicate among those causes. What it shows unambiguously is that a resident's state is the single strongest predictor of how much of their care is delivered by a temporary nurse.

There is a clear regional shape to the top of the table. New England and the Mid-Atlantic — Vermont, New Hampshire, Maine, New Jersey, New York, Pennsylvania — cluster among the most agency-reliant, alongside a few thin-labor-market frontier states (Alaska, New Mexico, North Dakota). The lowest-reliance states form their own block across the South and lower-cost Sun Belt: Alabama, Florida, Arkansas, Texas. The pattern is consistent enough to be structural rather than incidental.

Intensity is not the same as volume

One nuance is worth drawing out, because it is the kind of thing a single ranked column hides. Vermont has the highest agency intensity — a quarter of its hours — but with only 34 nursing homes, it accounts for just 0.9% of all the agency hours billed in the country. The states that dominate the national agency bill are the large, moderately-reliant ones. New York alone, at 10.8% reliance across 594 homes, supplies 13.8% of every agency hour worked in a US nursing home. Pennsylvania adds another 12.0%, and New Jersey 6.9%. Three states — New York, Pennsylvania, New Jersey — account for 32.7% of the national agency total between them.

So there are two different stories in the same column. If the question is "where is a resident most likely to be cared for by a temp," the answer is Vermont. If the question is "where is the money for agency nursing actually being spent," the answer is the New York–Pennsylvania–New Jersey corridor. Both are true, and conflating them is the most common way this kind of figure gets misread.

For the per-facility detail behind any state, the full daily record is published with source provenance on the PBJ staffing dataset page, and the facility-quality context for nursing homes more broadly lives in the Care Compare nursing-home module.

The concentration nobody sees in the average

The 5.98% national figure is an average, and averages hide tails. Group the country's homes by how much of their own nurse staffing they buy from agencies, and the distribution turns out to be sharply bimodal.

Two facts in this table do the work. First, 41.5% of all nursing homes — 6,037 of 14,537 — bought no agency hours at all in the quarter. They ran entirely on their own employees. The median home's agency share was essentially zero (0.50%). For most of the system, agency staffing simply is not a feature of how it operates.

Second, the reliance that exists is highly concentrated. The 854 homes that bought a quarter or more of their nurse hours from agencies — just 5.9% of all facilities — accounted for 41.5% of every agency hour billed nationwide. And at the extreme, 86 facilities ran on majority-agency staff: more than half of all their nurse hours, for an entire quarter, came from temporary workers. Nineteen of those topped 75%. These are not homes that covered a bad weekend when a nurse called out. They are homes whose ordinary operating model, for three months straight, was built on rented coverage.

Per our research doctrine, we do not name individual facilities or attach these counts to any provider profile. CMS publishes the underlying file at the facility level for anyone who wishes to look; our role is to report the aggregate pattern, not to rate, warn against, or endorse any specific home.

Why the continuity gap matters

The reason an agency hour is not interchangeable with an employee hour has little to do with the individual nurse, who is often excellent, and everything to do with continuity. A permanent nurse knows which resident is normally chatty and is now withdrawn, which one's gait changed this week, which family member to call. That accumulated, undocumented knowledge is part of what catches a deteriorating resident early. A nurse arriving for a single agency shift in an unfamiliar building, however skilled, starts without it.

The staffing literature has consistently associated heavy agency reliance with weaker continuity of care, higher turnover, and the operational strain that produces both. We are careful not to overstate this: the Payroll-Based Journal records hours, not outcomes, and this study draws no causal line from an agency hour to any resident's harm. What the file establishes is the input — how much of the nursing workforce in a given home, state, or the country as a whole is temporary rather than permanent — and it establishes it precisely, from payroll, for every certified home in the nation.

A note on direct-care registered nurses

Readers of our companion study on zero-RN days will recognize the registered-nurse direct-care column, where the agency share runs higher than the all-roles figure — about 7.3% of RN direct-care hours, against 5.78% across all three RN job codes combined and 5.98% across the whole nurse workforce. The difference is a denominator choice, not a contradiction: the narrower the role you measure, the more agency-dependent it tends to look, because the licensed bedside roles are the hardest to staff permanently. We report the all-roles figure as the headline here because it answers the broadest version of the question — what share of nursing-home nursing, of every kind, is now temporary.

What this study does, and does not, show

This is a descriptive study built on a single quarter of a single federal file. Its limits matter as much as its findings.

• It is one quarter. April–June 2025 is a single snapshot, captured as of 2025-06-30. Agency reliance moves seasonally and with the local labor market; a home's share here may not be its share today. The figures will shift as CMS publishes later quarters.
• It measures hours, not quality. Agency share is a staffing input. This study does not measure resident outcomes and makes no claim that agency reliance caused any specific harm.
• It is not risk-adjusted. Shares are raw ratios, not adjusted for resident acuity, case mix, facility size, or local wage levels. State differences reflect those factors alongside staffing decisions, and the file cannot separate them.
• It depends on facility self-report. PBJ is payroll-derived and auditable, which makes it far stronger than the old self-reported survey week, but the employee-versus-contract split is still coded by the facilities themselves. Mis-coding an agency nurse as an employee, or the reverse, would shift these shares.
• No facility-level claims. State and national aggregates are research outputs. We attach nothing to any provider profile and rate, endorse, or warn against no individual home.

None of these limits change the central, re-checkable fact: in the most recent published quarter, roughly one nurse hour in seventeen across America's nursing homes was bought from a staffing agency, and that reliance was concentrated — by role, by state, and by facility — in patterns too sharp to be accidental.

Medical review

Reviewed by Jennifer Montecillo, MD, medical reviewer. Non-practicing medical reviewer. This study reports staffing figures drawn from a public CMS payroll file; it does not assess, rate, or make clinical claims about any individual facility or resident, and it does not offer medical advice.

Reproduce this

Every figure above traces to the CMS Payroll-Based Journal 2025 Q2 snapshot, the immutable quarter CMS published for the period 2025-04-01 through 2025-06-30. The exact SQL that regenerates each headline figure ships with this study in the reproducibility block below, and the daily source file is documented on the PBJ staffing dataset page. Re-check any figure here against the source on our re-check surface; the broader provenance posture for every source family we ingest is catalogued at /sources, and the research-desk methodology standard is at /methodology.

For readers using this data operationally, our exclusion and sanction-list monitoring and the standalone staffing brand hub build on the same nursing-home data graph. For adjacent analysis, see our companion studies on zero-RN days, on how concentrated nursing-home ownership has become, and on the rising size of nursing-home fines.

Further reading and primary sources

• CMS Payroll-Based Journal Daily Nurse Staffing — the source dataset, at data.cms.gov.
• CMS nursing-home provider data topics — facility-level extracts, at data.cms.gov.
• The 2024 minimum-staffing final rule (since rescinded) — the regulatory text, at federalregister.gov.
• Independent policy analysis of nursing-home staffing and agency reliance — KFF and MACPAC both track the staffing debate and its evidence base.

The underlying CMS data is a U.S. Government Work in the public domain. Reuse of this study is permitted with attribution to Fonteum Research and a link back to this page.