3G Healthcare Real Estate · SNFRadar Intelligence
An independent study · September 2026 · Corrected September 14
The Rented Shift
Agency nursing hours and subsequent changes of ownership in skilled nursing. A retrospective comparison of federal staffing and Medicare enrollment records.
The short version
Seven findings, if you read nothing else
- Higher agency use was associated with more subsequent Medicare ownership changes. The recorded two-year rate was 7.46% below 2% agency and 11.79% at 15% or more: 1.6 times the unadjusted rate. These are facility-year observations, not counts of properties sold.
- The comparison varies by year. Giving both groups the same observation-year weights reduces the ratio to 1.35. That adjusts for year composition only; it does not establish causation or an individual facility’s sale probability.
- Across eight pooled agency bands, the recorded rate ranges from 6.98% in the rounded-zero band to 14.87% at 40% or more. The pooled pattern does not demonstrate a threshold or a uniform relationship within every year.
- Heavy agency use appears across nursing roles. The recorded ownership-change rates were 10.96% for heavy registered-nurse agency use, 11.29% for practical nurses and 11.58% for aides. Those groups overlap.
- Rating and occupancy comparisons are descriptive. We show both agency groups within broad bands, including separate unknown categories. Those cuts do not hold all other facility or market characteristics constant.
- In the latest observed trailing year, ending March 2026, 1,580 of 14,095 qualifying facilities, 11.2%, averaged at least 15% agency. This is the latest published staffing window, not a September headcount.
- Agency use fell in both descriptive before-and-after groups. Heavy users with a recorded change went from 28.43% to 17.07%; heavy users without a change in the held record went from 26.44% to 20.46%. The comparison cannot tell us how much a new owner caused.
In plain terms. We observed nursing homes that reported a full year of staffing in 2018, 2019, 2021, 2022 and 2023 (2020 is left out because reporting was optional for part of it) and followed each building for the next two years. Heavy agency use means at least 15% of a building’s nursing hours were worked by agency staff; light means under 2%. Buildings that leaned heavily on agency nurses went on to have a Medicare ownership change recorded about 1.6 times as often as buildings that barely used them, and about 1.35 times as often once the years are lined up. That is a pattern in the records. It does not prove that agency use causes a sale, an ownership change in the Medicare file is not the same as a building being sold, and it says nothing about any one building. The staffing figures run through the quarter ending March 2026, the latest CMS has published; that is about six months old, which is normal for these files. The words used above are explained under the numbers below.
In plain English
How to read this study
This is a research document, so it uses some technical words. Here is what each one means, in the order you are likely to meet them.
- Agency nursing hours
- Hours worked by nurses and aides supplied by a staffing agency rather than employed by the building. Federal filings call them contract hours. A building’s agency use is the share of all its nursing hours that agency staff worked.
- Heavy and light agency use
- Heavy means at least 15% of nursing hours came from agency staff. Light means under 2%. These are our cut points for comparison, not an official standard.
- Medicare ownership change
- A record that the Medicare enrollment for a nursing home passed to a different owner. It is a paperwork event in a federal file. It does not always mean the real estate was sold, and a building can be sold without one appearing here. Throughout this study, an ownership change means this record, not a deed.
- Facility-year
- One building, observed in one year. The main comparison covers 68,896 facility-years drawn from 15,619 buildings, so most buildings appear several times, once for each year we could observe them. It is a count of observations, not of buildings and not of sales.
- Recorded two-year rate
- Out of all the facility-years in a group, the share that had a Medicare ownership change recorded within the following 24 months.
- Unadjusted
- A straight comparison of two groups with nothing else held equal. The 1.6 times headline is the heavy group’s rate divided by the light group’s rate, and nothing more.
- Year-adjusted, or standardized
- The same comparison after giving both groups the same mix of years. Some years saw more ownership changes than others, and the heavy group had more of its observations in those years, so lining the years up narrows the gap to 1.35 times.
- Association, not cause
- We looked back at records that already exist. Nobody assigned buildings to use agency staff, so the study can show that heavy agency use and later ownership changes go together more often than light use does. It cannot show that one caused the other, and it cannot predict what will happen to any single building.
- Descriptive
- A comparison that reports what the records show for a group, without trying to hold every other difference between buildings constant.
- Denominator
- The number a rate is measured out of. A rate built on a handful of buildings is noise, so the state table withholds any rate with fewer than 30 observations and shows the count instead.
- Payroll-Based Journal
- The federal system in which every certified nursing home reports its daily nursing hours from payroll, separating employees from agency staff. It is the source of every staffing figure here.
- Trailing four quarters
- The twelve months ending on a given date. A building’s agency share for a year is the average of its four quarterly shares.
- Deed overlap
- A separate check against our own property-transaction records, asking how often a Medicare ownership change lines up in time with a recorded property sale. It is a rough cross-check, and none of the headline figures depend on it.
The rest of the study
Every finding, every chart, and the method behind each figure.
The body of the study opens with an account. It is free, it takes one text message, and it opens every state report we publish along with it.
The state table below stays open to everyone, account or not, and so does the CSV.
The open state table
Agency reliance and change of ownership by state
In plain terms. Each row is one state. Heavy share is the portion of that state’s observed building-years that were heavy agency users. The two change rates show how often heavy and light users went on to have a recorded ownership change, and the observation counts beside them tell you how much weight each rate can bear. The latest columns describe the most recent published year only.
Historical columns use 2018, 2019, 2021, 2022 and 2023 facility-years, each followed for 24 months. Latest columns use the four quarters ending March 2026. Heavy means at least 15%; light means under 2%. A rate is withheld below 30 observations; the count remains visible. Latest state shares require at least 30 facilities. Rates are descriptive Medicare enrollment outcomes, not property sale rates.
| State | Distinct facilities | Facility-years | Heavy share | Heavy observations | Heavy change rate | Light observations | Light change rate | Latest facilities | Latest heavy | Latest heavy share |
|---|---|---|---|---|---|---|---|---|---|---|
| AK | 15 | 46 | 17.4% | 8 | n/a | 22 | n/a | 11 | 3 | n/a |
| AL | 225 | 993 | 3.8% | 38 | 5.3% | 812 | 4.8% | 220 | 3 | 1.4% |
| AR | 230 | 961 | 1.7% | 16 | n/a | 857 | 11.2% | 211 | 5 | 2.4% |
| AZ | 150 | 661 | 8.5% | 56 | 10.7% | 382 | 7.3% | 133 | 8 | 6.0% |
| CA | 1,188 | 5,244 | 4.7% | 249 | 12.0% | 3,747 | 8.2% | 1,042 | 42 | 4.0% |
| CO | 230 | 1,020 | 14.3% | 146 | 24.7% | 489 | 9.6% | 198 | 25 | 12.6% |
| CT | 214 | 927 | 5.8% | 54 | 1.9% | 568 | 6.5% | 188 | 24 | 12.8% |
| DC | 18 | 71 | 4.2% | 3 | n/a | 53 | 5.7% | 15 | 3 | n/a |
| DE | 45 | 207 | 14.0% | 29 | n/a | 126 | 6.3% | 42 | 3 | 7.1% |
| FL | 722 | 3,246 | 11.1% | 360 | 13.9% | 1,970 | 9.8% | 679 | 9 | 1.3% |
| GA | 361 | 1,656 | 12.8% | 212 | 10.8% | 983 | 5.9% | 346 | 16 | 4.6% |
| HI | 44 | 185 | 7.0% | 13 | n/a | 110 | 2.7% | 42 | 9 | 21.4% |
| IA | 445 | 1,986 | 16.3% | 323 | 2.8% | 884 | 7.0% | 377 | 31 | 8.2% |
| ID | 80 | 360 | 5.0% | 18 | n/a | 219 | 7.8% | 76 | 8 | 10.5% |
| IL | 731 | 3,174 | 12.7% | 402 | 16.4% | 1,909 | 11.8% | 652 | 66 | 10.1% |
| IN | 547 | 2,495 | 11.1% | 278 | 6.5% | 1,560 | 3.9% | 494 | 29 | 5.9% |
| KS | 330 | 1,440 | 14.6% | 210 | 3.3% | 773 | 6.6% | 284 | 36 | 12.7% |
| KY | 284 | 1,275 | 12.2% | 156 | 12.2% | 809 | 8.3% | 262 | 17 | 6.5% |
| LA | 271 | 1,194 | 8.0% | 96 | 21.9% | 767 | 6.8% | 248 | 13 | 5.2% |
| MA | 393 | 1,706 | 15.2% | 259 | 13.9% | 748 | 5.3% | 331 | 38 | 11.5% |
| MD | 225 | 1,053 | 18.0% | 190 | 16.8% | 529 | 14.0% | 218 | 46 | 21.1% |
| ME | 98 | 421 | 35.4% | 149 | 0.7% | 165 | 0.0% | 72 | 22 | 30.6% |
| MI | 444 | 1,996 | 7.2% | 144 | 7.6% | 1,430 | 3.4% | 398 | 20 | 5.0% |
| MN | 373 | 1,631 | 9.3% | 152 | 8.6% | 999 | 1.7% | 320 | 61 | 19.1% |
| MO | 525 | 2,189 | 10.2% | 224 | 14.7% | 1,497 | 5.6% | 463 | 46 | 9.9% |
| MS | 208 | 930 | 13.1% | 122 | 1.6% | 605 | 2.8% | 194 | 16 | 8.2% |
| MT | 74 | 302 | 34.8% | 105 | 23.8% | 108 | 8.3% | 55 | 16 | 29.1% |
| NC | 419 | 1,920 | 21.4% | 411 | 18.7% | 1,012 | 7.4% | 401 | 94 | 23.4% |
| ND | 80 | 355 | 25.1% | 89 | 0.0% | 134 | 3.0% | 67 | 26 | 38.8% |
| NE | 204 | 857 | 20.5% | 176 | 8.0% | 363 | 3.6% | 168 | 39 | 23.2% |
| NH | 73 | 346 | 18.8% | 65 | 7.7% | 107 | 4.7% | 69 | 18 | 26.1% |
| NJ | 366 | 1,655 | 18.7% | 310 | 13.2% | 754 | 10.3% | 343 | 92 | 26.8% |
| NM | 75 | 331 | 19.0% | 63 | 19.0% | 129 | 12.4% | 66 | 18 | 27.3% |
| NV | 64 | 276 | 9.1% | 25 | n/a | 179 | 4.5% | 62 | 5 | 8.1% |
| NY | 618 | 2,794 | 26.7% | 745 | 0.9% | 1,036 | 1.4% | 579 | 112 | 19.3% |
| OH | 985 | 4,354 | 13.3% | 581 | 7.2% | 2,495 | 5.2% | 910 | 74 | 8.1% |
| OK | 298 | 1,201 | 5.9% | 71 | 18.3% | 898 | 9.7% | 256 | 8 | 3.1% |
| OR | 134 | 578 | 15.2% | 88 | 27.3% | 242 | 5.8% | 121 | 24 | 19.8% |
| PA | 697 | 3,180 | 23.3% | 742 | 17.1% | 1,283 | 6.3% | 635 | 170 | 26.8% |
| PR | 5 | 15 | 0.0% | 0 | n/a | 12 | n/a | 7 | 0 | n/a |
| RI | 82 | 356 | 13.5% | 48 | 18.8% | 201 | 4.0% | 71 | 8 | 11.3% |
| SC | 193 | 878 | 16.7% | 147 | 18.4% | 480 | 6.0% | 180 | 32 | 17.8% |
| SD | 108 | 442 | 16.3% | 72 | 19.4% | 200 | 6.5% | 90 | 22 | 24.4% |
| TN | 318 | 1,472 | 11.3% | 167 | 6.0% | 1,041 | 6.0% | 294 | 23 | 7.8% |
| TX | 1,254 | 5,336 | 7.6% | 405 | 16.5% | 3,687 | 13.0% | 1,136 | 50 | 4.4% |
| UT | 104 | 467 | 15.0% | 70 | 7.1% | 277 | 3.6% | 95 | 5 | 5.3% |
| VA | 293 | 1,340 | 15.0% | 201 | 16.9% | 749 | 10.9% | 285 | 34 | 11.9% |
| VT | 36 | 161 | 37.9% | 61 | 3.3% | 37 | 0.0% | 33 | 22 | 66.7% |
| WA | 212 | 913 | 13.5% | 123 | 22.0% | 458 | 8.7% | 184 | 18 | 9.8% |
| WI | 370 | 1,572 | 11.2% | 176 | 7.4% | 867 | 5.5% | 318 | 57 | 17.9% |
| WV | 125 | 570 | 13.2% | 75 | 20.0% | 314 | 14.0% | 120 | 9 | 7.5% |
| WY | 36 | 158 | 12.0% | 19 | n/a | 97 | 8.2% | 34 | 5 | 14.7% |
Limits and source dates
What these results do not establish
In plain terms. This study shows a pattern in federal records. It does not say why a building uses agency staff, it does not predict whether any one building will be sold, and it cannot see ownership changes that were never filed with Medicare. Both files are published with a lag, so the most recent months are incomplete. The figures were rebuilt on September 14, 2026 after two missing staffing files were recovered; the original edition is kept for anyone who cited it.
Agency use may respond to labor markets, staffing needs, management choices, financial circumstances or an anticipated transition. This study does not identify a cause or a seller’s motive. The headline is an unadjusted group comparison, not an individual sale forecast. We have not estimated clustered uncertainty, performed out-of-sample prediction or established independence between repeated facilities and overlapping follow-up windows.
A change effective one day after an observation can count alongside one nearly two years later. Staffing publication and enrollment updates both lag events. Measuring exposure before an effective date does not prove a year of usable advance notice. An absent event means no qualifying event in the held record, not proof ownership never changed.
The staffing snapshot ends 2026Q1; the held enrollment record ends 2026-02-01. Its monthly tail below is an audit aid, not evidence that transaction activity truly fell at the end of the file. Data recovered and results rebuilt September 14, 2026; the original September 2 edition is preserved.
Public sources: CMS Payroll-Based Journal nurse staffing, CMS skilled nursing change of ownership, CMS nursing-home provider information for rating and occupancy, and the June 25, 2020 reporting memorandum. The supplemental deed-overlap check uses proprietary transaction metadata held by 3G; it does not determine the federal panel’s outcomes.
| Effective month | Recorded enrollment changes |
|---|---|
| 2024-02 | 41 |
| 2024-03 | 46 |
| 2024-04 | 40 |
| 2024-05 | 46 |
| 2024-06 | 65 |
| 2024-07 | 41 |
| 2024-08 | 100 |
| 2024-09 | 68 |
| 2024-10 | 45 |
| 2024-11 | 57 |
| 2024-12 | 108 |
| 2025-01 | 53 |
| 2025-02 | 41 |
| 2025-03 | 25 |
| 2025-04 | 30 |
| 2025-05 | 42 |
| 2025-06 | 71 |
| 2025-07 | 23 |
| 2025-08 | 28 |
| 2025-09 | 26 |
| 2025-10 | 25 |
| 2025-11 | 17 |
| 2025-12 | 14 |
| 2026-01 | 2 |
| 2026-02 | 1 |
For journalists and researchers
Use the corrected figures with attribution
The full state table and its CSV are open. We can provide the aggregate figures and methodology behind each comparison. Published versions are dated; this correction supersedes the September 2 figures.
Interviews and comment: Stan Klos III, [email protected]. Data and methodology: Christian Patrick, [email protected].
Cadence. Staffing figures are reviewed against each new Payroll-Based Journal release. Any published revision will be dated so readers can distinguish it from the version they cited.
About
Who published this
3G Healthcare Real Estate exclusively represents sellers of skilled nursing facilities. SNFRadar is its research operation. We publish aggregate research to help operators and readers understand the records available to them.
3G Healthcare Real Estate · Exclusively SNF Sell-Side Advisors · SNFRadar Intelligence
The Rented Shift · Corrected September 14, 2026 · Free to cite with attribution
Public CMS staffing and enrollment records; supplemental proprietary deed-overlap metadata. No individual facility, owner or client is identified.