Nursing homes book $15.94 billion a year to companies their own owners control
Across 10,948 US nursing homes that file a Medicare cost report with the figure, the median home books 10.4% of its operating expenses — not its revenue — to organisations related to it by common ownership or control. The spread between states is more than threefold.
This is legal, disclosed, and ordinary industry structure. A nursing home may lease its building from a company its owners also own, or buy management, therapy, staffing or pharmacy services from an affiliate. Medicare requires those arrangements to be reported precisely because they are common, and the figures below are the operators’ own filings, not our estimates. Nothing on this page says a home did anything improper. What it says is how much money moves this way, where, and what it does and does not track.
The scale
- $15.94 billion booked to related parties, across 10,948 of 13,910 homes with a cost report (78.7%). This is a floor, not a total — the 2,962 homes without the figure are excluded, not counted as zero.
- Median $1,009,830 per home.
- As a share of operating expense: median 10.4%, 75th percentile 16.3%, 90th percentile 22.7%.
- 743 homes (6.8%) book a quarter or more of everything they spend to affiliated companies.
It depends heavily on the state
Each figure below is the median individual home in that state, not the state’s total spending divided by its total expenses. A state median of 20% means the typical home there books a fifth of its spending to affiliates — it does not mean a fifth of all money in the state does. 40 states have at least 50 homes with the figure; smaller states are omitted rather than shown on thin counts.
| State | Median share of operating expense | Homes |
|---|---|---|
| AR | 20.2% | 189 |
| NC | 17.3% | 357 |
| IN | 14.8% | 481 |
| MO | 13.9% | 410 |
| NJ | 13.7% | 287 |
| KY | 13.1% | 201 |
| TN | 13.0% | 249 |
| LA | 12.8% | 229 |
| AL | 12.6% | 182 |
| GA | 12.6% | 277 |
| MI | 12.3% | 363 |
| IL | 11.6% | 585 |
| AZ | 11.5% | 110 |
| NY | 11.3% | 383 |
| RI | 11.0% | 57 |
| UT | 11.0% | 75 |
| NM | 10.7% | 55 |
| WV | 10.6% | 90 |
| FL | 10.6% | 571 |
| ID | 10.5% | 66 |
| CT | 10.5% | 175 |
| MD | 10.2% | 190 |
| VA | 10.1% | 234 |
| CO | 9.7% | 145 |
| OH | 9.6% | 799 |
| MA | 9.3% | 296 |
| MS | 9.2% | 161 |
| KS | 9.1% | 207 |
| OK | 9.0% | 175 |
| PA | 8.6% | 497 |
| NE | 8.5% | 100 |
| MN | 8.2% | 189 |
| OR | 8.0% | 103 |
| SC | 8.0% | 160 |
| IA | 7.9% | 270 |
| TX | 7.3% | 390 |
| WA | 7.2% | 153 |
| CA | 7.1% | 902 |
| ME | 6.4% | 64 |
| WI | 6.2% | 212 |
What this doesn’t tell you
It does not track care quality. Sorted by the CMS overall star rating, related-party share is essentially flat — a spread of 0.8 percentage points between the highest and lowest star median, across 10,691 rated homes. The share of homes at 25% or more is flat too. If you expected homes routing more money to affiliates to be rated worse, that is not what these filings show, and we would rather publish that than leave it out.
| CMS overall rating | Median share | Share of homes at ≥25% | Homes |
|---|---|---|---|
| 1 | 10.2% | 7.0% | 2,284 |
| 2 | 10.4% | 6.5% | 2,275 |
| 3 | 10.3% | 6.8% | 2,133 |
| 4 | 10.9% | 6.4% | 2,043 |
| 5 | 10.1% | 7.3% | 1,956 |
One exception, at the top of the staffing scale
Start with what does not move. Between one and four stars on the payroll-based staffing measure, the median share barely changes — 11.2% to 9.9% — and the difference appears only at the top of the scale, in the smallest group. This is not a steady relationship where better-staffed homes send less money to affiliates at every level. It is a single step at five stars.
| Staffing rating | Median share | 75th percentile | Homes |
|---|---|---|---|
| 1 | 11.2% | 17.1% | 1,831 |
| 2 | 11.3% | 17.7% | 2,573 |
| 3 | 10.8% | 16.3% | 2,885 |
| 4 | 9.9% | 15.7% | 2,464 |
| 5 | 7.1% | 11.6% | 923 |
And it is not the same everywhere. Of the 24 states with at least 15 five-star-staffed homes to compare, the step appears in 17, is absent in 4, and runs the other way in 3. Any summary of this finding that leaves out the states where it fails is overstating it.
Two checks it does survive. It is not simply a description of who owns what: the step persists when the comparison is restricted to for-profit homes only, to chain-owned homes only, to both at once, and to non-profits only — though it shrinks as the comparison narrows, so ownership does account for part of the raw difference. And it is not an artefact of a small group’s median: the five-star group sits lower at the 75th percentile as well, so the whole distribution shifts.
| Restricted to | Median, 1–4 stars | Median, 5 stars | Difference (pp) | Homes |
|---|---|---|---|---|
| All homes | 10.8% | 7.1% | +3.7 | 10,676 |
| For-profit only | 11.2% | 8.6% | +2.5 | 8,987 |
| Chain-owned only | 11.3% | 7.6% | +3.7 | 8,596 |
| For-profit and chain-owned | 11.5% | 9.6% | +1.9 | 7,537 |
| Non-profit only | 8.0% | 6.4% | +1.6 | 1,419 |
Where it fails looks like a ceiling effect, and we tested that rather than assuming it. The size of a state’s step tracks how much related-party spending that state has to begin with (Spearman ρ = +0.81, Pearson r = +0.80, across 24 states). Every state where the step is absent or reversed has a baseline below 9.7% — there is little room for a decline where the typical home already books little. That is an explanation for our own disconfirming cases, so it needed evidence rather than assertion; had those states been scattered across the range we would have dropped the explanation and reported the count alone.
| State | Median, 1–4 stars | Median, 5 stars | Difference (pp) | Homes 1–4 | Homes 5 |
|---|---|---|---|---|---|
| MN | 8.8% | 6.9% | +1.9 | 108 | 79 |
| CA | 7.3% | 6.7% | +0.7 | 816 | 77 |
| PA | 9.2% | 8.3% | +1.0 | 429 | 58 |
| MI | 13.0% | 7.0% | +5.9 | 299 | 56 |
| FL | 11.2% | 9.1% | +2.1 | 508 | 54 |
| IA | 8.3% | 6.1% | +2.2 | 208 | 41 |
| KS | 9.0% | 10.9% | -2.0 | 164 | 33 |
| MS | 8.9% | 11.0% | -2.2 | 126 | 33 |
| AL | 13.2% | 7.8% | +5.3 | 149 | 32 |
| WI | 6.5% | 5.6% | +0.9 | 175 | 32 |
| IL | 12.1% | 6.1% | +6.0 | 528 | 31 |
| MA | 9.6% | 7.0% | +2.6 | 262 | 28 |
| OR | 8.7% | 5.8% | +2.9 | 76 | 26 |
| NC | 17.7% | 8.9% | +8.8 | 327 | 24 |
| WA | 6.6% | 10.6% | -4.0 | 128 | 23 |
| NJ | 14.0% | 8.1% | +6.0 | 262 | 21 |
| ME | 11.6% | 3.2% | +8.3 | 37 | 20 |
| CT | 11.0% | 6.0% | +4.9 | 147 | 19 |
| OH | 9.7% | 8.7% | +1.0 | 766 | 19 |
| MD | 11.7% | 6.4% | +5.3 | 171 | 17 |
| TX | 7.6% | 4.6% | +3.1 | 358 | 17 |
| VA | 11.6% | 6.0% | +5.6 | 215 | 17 |
| CO | 9.9% | 7.7% | +2.2 | 127 | 16 |
| NE | 12.5% | 5.0% | +7.5 | 79 | 16 |
Most homes book more than Medicare allows
Medicare does not accept a related-party charge at face value. Because a payment to an affiliate is partly a payment to yourself, the rules allow such costs only at the related organisation’s own cost, and the cost report carries both figures: what the home booked, and what is allowable. The difference is an adjustment.
- 6,579 homes (58.7%) booked more than the allowable amount — $3.06 billion above the ceiling in total.
- 3,946 (35.2%) booked less than allowable, totalling $1.51 billion.
- Net adjustment across 11,201 filings: −1.55 billion removed from claimed costs.
The adjustment is a routine part of cost reporting, and an amount above the ceiling is not evidence of wrongdoing — it is what the worksheet exists to correct. It does show that for most homes, the amount first booked to affiliates exceeded what Medicare would ultimately allow.
Method
- Source. Medicare cost reports (HCRIS, form CMS-2540-10), Worksheet A-8-1, column 5 — “Amount Included in Wkst. A., col. 5”, what the provider booked in its own expenses. This is easy to confuse with column 4, “Amount Allowable In Cost”, which is the Medicare ceiling rather than the amount booked. Every figure above except the adjustment section uses column 5.
- Denominator. Share of operating expense, the same basis shown on each facility page, so the two reconcile. Shares of revenue would be lower and are not used here.
- One filing per home. Where a facility has filed more than once we use the most recent — latest fiscal-year end, then latest processing date, then highest report number. Figures are therefore not a single fiscal year: 9,439 ending in 2023 and 1,509 ending in 2024.
- Coverage. 10,948 of 13,910 homes (78.7%) have the figure. Totals are floors; homes without it are excluded rather than treated as zero.
- Thresholds. States are shown at 50+ homes; the within-state staffing comparison requires 15+ five-star homes and 30+ homes at one to four stars. These are stated so you can see what was left out.
- Reproducible. Every number on this page is computed by
scripts/report_related_party.pyin our repository and can be re-derived from the public HCRIS release. Our full methodology covers the rest of the site. - Corrections. If a figure here is wrong, or a home’s cost report is wrong, see our corrections policy — it sets out which of those we can fix and where the other belongs.