Where can I get free health care REIT data? Both halves of the industry, delivered as rows

The query reads like an access question and lands as a coverage question. Nobody typing it wants a link collection; they want to know whether the thing they are building - a senior-housing absorption model, a tenant-revenue screen, a sector factor - has a supply at all, or only charts and press releases. It does. A health care REIT collects rent from hospitals, medical office buildings, senior housing and skilled-nursing facilities, so the data splits into two families that rarely share a shelf: what the tenants' healthcare business is actually doing, and what the listed equity is doing.

[Datadory](/) carries both under one roof: 14 pooled datasets for this industry - eight primary records plus six cross-listed neighbors - sorted into five working layers.

• Delivery economics - CMS Data Hub - Medicare & Medicaid Datasets, 159 cataloged program datasets, the slice's only quality-10 record.
• Facility inventory - CMS Provider Data Catalog, 236 provider files from hospitals to hospices.
• Pricing and variation - hospital charges by DRG and APC beside Dartmouth Atlas regional rates.
• Demand demographics - two CDC collections whose mortality series reach 1915.
• Equity and benchmark - per-ticker quote records and the FTSE Nareit index stack.

One record stands apart from all of them at the grain level, and section six gets to it. The rest of this page works the layers in order, with the verified numbers each product ships.

The Centers for Medicare & Medicaid Services publish probably the single richest record of American health care delivery, and CMS Data Hub - Medicare & Medicaid Datasets packages it as one maintained feed: 159 cataloged datasets spanning Medicare (153 tagged), Medicaid (8), the Children's Health Insurance Program (6) and the Health Insurance Marketplace (2), organized into nine families from provider characteristics (34 datasets) through utilization-and-payment summaries (31) to beneficiary enrollment statistics (10). The provider-summary family covers the 2013-2024 service years, and it scores 10 of 10 on Datadory's field-documentation rubric - the highest mark in this slice.

Two cuts of one hospital show why the record earns that score:

Rndrng_Prvdr_CCN            010001
Rndrng_Prvdr_Org_Name       Southeast Health Medical Center
Rndrng_Prvdr_City           Dothan
Rndrng_Prvdr_State_Abrvtn   AL
DRG_Cd                      003
Tot_Dschrgs                 11
Avg_Submtd_Cvrd_Chrg        738478.64
Avg_Tot_Pymt_Amt            103236.27
Avg_Mdcr_Pymt_Amt            91218.18

Rndrng_Prvdr_CCN is the load-bearing column: the CMS Certification Number keys every provider-level file the agency publishes, so a facility table built on it absorbs enrollment, spending and quality families without remapping identifiers. RUCA descriptions ride on the row, which settles metro-versus-rural slicing before it starts, and the demographic panel (Bene_Avg_Age 75.47 at this facility, Bene_Dual_Cnt among them) is the cleanest demand measure a senior-housing or skilled-nursing model can ask for. Enrollment and spending trends here are the macro driver behind every downstream tenant-revenue assumption.

Inside sit the files senior-housing and medical-outpatient analysts argue about. The nursing home Provider Information file alone runs roughly 15,000 rows by 100+ columns, and one row reads like this:

ccn                      015009
provider_name            BURNS NURSING HOME, INC.
city_town                RUSSELLVILLE
state                    AL
ownership_type           For profit - Corporation
number_of_certified_beds 57
avg_residents_per_day    51.6
overall_rating           2
latitude                 34.5149
longitude                -87.7360

Three properties make this the operator-mapping record. First, the four rating columns decompose reputation into inspectable parts - overall_rating moves when health_inspection_rating, staffing_rating or qm_rating move, so a submarket thesis can name its lever. Second, ownership and chain fields reveal whether a target portfolio's buildings sit with one operator before underwriting begins. Third, Datadory retains every delivered snapshot, which is how star-rating and staffing histories accrue into a proper panel instead of overwriting themselves - the difference between a current-state screenshot and a rating history.

Because ccn keys this catalog too, the facility map joins straight onto the Data Hub's service-line dollars and the charge record below it. One identifier, three layers, no reconciliation project.

Four grains, arriving together - which is the whole integration problem in one glance:

# facility row - CMS Provider Data Catalog, nursing home file
ccn                     : 015009      beds     : 57
provider_name           : BURNS NURSING HOME, INC.
overall_rating          : 2           census   : 51.6 residents/day

# service-line dollar row - CMS charge data, inpatient
Rndrng_Prvdr_CCN        : 010001      DRG_Cd   : 003
Tot_Dschrgs             : 11
Avg_Submtd_Cvrd_Chrg    : 738478.64   Avg_Tot_Pymt_Amt : 103236.27

# equity row - quote card, largest name in the cohort
symbol                  : WELL        price    : 237.40
trailing_pe             : 105.04      fwd_dividend_and_yield : 3.40 (1.43%)

Each row keeps its own unit, period and provenance stamp - dollars per discharge here, a percent-of-yield there, a market-close stamp on every quote cell - so the layers stack without lying to each other. That discipline is what lets one table hold a facility's star rating, the price its tenant charged per DRG, the sector's cash-flow conversion and the multiple the market pays for it. Assembling those grains from separate shelves means reconciling schemas that were never meant to meet; delivered together, the join keys arrive pre-agreed - CCN, DRG, period, ticker.

One row per de-identified stay, in the shape it arrives:

KEY_NIS      : <encounter key>          HOSP_NIS    : <masked hospital id>
AGE          : <age in years>           FEMALE      : <sex indicator>
DXCCSR1..N   : <all-listed diagnoses -> CCSR categories>
PAY1         : <1 Medicare | 2 Medicaid | 3 private | 4 self-pay | 5 no charge>
TOTCHG       : <total charges for the stay, nominal dollars>
DISPUNIFORM  : <routine | transfer | home health | died>
DISCWT       : <discharge weight expanding the sample to national totals>

PAY1 is why all-payer questions land here: the commercially insured and uninsured stays that fee-for-service Medicare files never see sit on the same rows as everything else, which turns payer-mix diligence on a tenant operator into a group-by. HOSP_NIS keeps a masked facility identifier on every record, so encounters still aggregate back to buildings and join outward to the CCN-keyed layers above. DISPUNIFORM standardizes discharge destinations, and the Nationwide Readmissions Database tracks 30-day returns - the intensity measures post-acute underwriting runs on.

• Investors and quant researchers - screen the listed cohort on uniform quote fields, decompose the sector's total return into price and income against FTSE Nareit history reaching December 1971, and stress tenant revenue assumptions against charge-per-DRG trends from 2013 onward.
• REIT analysts and underwriters - attach star ratings, staffing hours and deficiency counts to every facility in a submarket before believing its rents, then read payer mix per referral market before underwriting revenue quality.
• Data scientists and ML engineers - join a century of vital statistics to today's facility roster on one normalized schema, where 1915 mortality cells and yesterday's closing prices land in the same DataFrame with types intact.
• Market researchers and consultants - size senior-housing and post-acute markets from HRR-level adjusted rates and citable federal counts, benchmarks no vendor recomputes.
• Sales and growth teams - rank hospitals by service-line mix - orthopedic-heavy, cardiac-heavy, oncology outliers - and time outreach to the operators actually expanding.
• Journalists and academics - quote inspection, staffing and penalty records tied to named facilities and published definitions rather than screenshot folklore.