The RAND Hospital Data is a streamlined collection of variables that make CMS hospital cost report data more accessible. We download raw CMS cost report data submitted by hospitals and improve upon it, enabling users to solve logistical problems and answer complex questions more easily.

The RAND Hospital Data are structured as panel datasets; that is, each observation (row) in the data corresponds to one year of data for one level of observation (e.g., hospital, county, state, etc. - see “What does level of observation mean?” for more information).

Each dataset has observations for all available years of data between 1996 and the latest year of released hospital cost report data. For example, a hospital-level file downloaded from the May 1, 2023, release will contain one record for each hospital in each year the hospital submitted a cost report between 1996 and 2021 (See “How long does it take for the RAND Hospital Data to publish a complete year of CMS Hospital Cost Report data?” for further explanation).

No. The RAND Hospital Data contain a useful subset of variables, comprising approximately 1,500 fields selected directly from the hospital cost report, plus additional variables we derive for users. Counting all columns, lines, and sub-lines, Form 2552-10 (the full CMS hospital cost report) contains around 38,000 separate alphanumeric fields and 327,000 separate numeric fields.

See "What if there is a CMS cost report field I need that is not in the RAND Hospital Data?" if you are interested in a variable that is not included in the RAND Hospital Data.

The RAND Hospital Data files contain information on all hospitals in the United States and Puerto Rico that are required to submit CMS cost reports. These include short term, acute care hospitals; critical access hospitals; psychiatric hospitals; children's hospitals; rehabilitation facilities; and long-term care facilities.

The RAND Hospital Data are updated four times a year, on or around the first day of February, May, August, and November.

The RAND Hospital Data Team makes the current and all historical releases of the RAND Hospital Data available for subscriber download. The Release Date identifies a version specific to the date on which RAND published the dataset. For example, “2025_02_01” indicates the version of the RAND Hospital Data published on February 1, 2025.

The Level of Observation is the characteristic or set of characteristics that allows the user to uniquely identify one row in a dataset. For example, in a dataset with hospital and year as the level of observation, each record represents one hospital in one year. In a dataset with a geographical area and year as the level of observation (e.g., county, CBSA, state, or national), each record represents the summary of all hospitals contained in one geographic area in one year.

CMS requires all facilities to submit an annual cost report, but it allows each facility to select whatever cost reporting period it would like. In most cases, a facility submits cost reports using its fiscal year, which will typically run either from October 1 through September 30 (federal fiscal year), January 1 through December 31 (the calendar year), or July 1 through June 30. Facilities can and do have other cost reporting periods. Some hospitals will also change their cost reporting period in the middle of a year and have a cost reporting period shorter than a year.

There are many options in addressing the lack of uniformity in hospital cost reporting periods. To make comparisons across hospitals, it is useful to have the cost report data processed so that each record covers a standardized time period.

Subscribers to the RAND Hospital Data can choose datasets where cost report information has been standardized to one of three time periods, or they can choose to download unweighted data.

The three standardized periods:

• Calendar year: Each observation is standardized to January 1 - December 31.
• Federal fiscal year: Each observation is standardized to October 1 - September 30. For example, federal fiscal year 2019 spans October 1, 2018, through September 30, 2019.
• Hospital fiscal year: Each observation is the sum of all information included in cost reports with a start date during a given federal fiscal year. For example, if a hospital submits cost report 1 covering October 1, 2018 through December 31, 2018, and then submits cost report 2 covering January 1, 2019, through December 31, 2019, information from both reports will be summed into a single observation because both have a beginning date that falls within the federal fiscal year 2019.

In the hospital cost report time period datasets, the data are unweighted, and each record corresponds to one hospital cost reporting period. So, the start and end dates will differ by hospital and will not necessarily be one year long.

Core-Based Statistical Areas (CBSAs) are defined by the U.S. Census Bureau, and each one contains a single core county, an urbanized area, or an urban cluster with a population of at least 10,000 people. Each core county is combined with any adjacent counties that have a high degree of social and economic integration with the core, as measured through commuting ties with the core or other counties associated with the core.

CBSAs are each classified as either a metropolitan area (urbanized area with a population of least 50,000 people) or a micropolitan area (urbanized area with a 10,000-50,000 population). When processing and creating RAND Hospital Data, all rural counties within a state - i.e., not included in a metropolitan or micropolitan area - are grouped and coded as “XX999” where “XX” is the 2-character postal abbreviation for the state, such as “PA999” for Pennsylvania.

Subscribers can choose between data with OR without outliers corrected. To correct data outliers, RAND applies an algorithm that identifies numeric values that fall far outside the normal range of variation and replaces them with interpolated values. In general, the data is allowed a very wide range of variation before being corrected; the degree of variation is adjusted based on the degree of observed variation within a given hospital over time (hospitals that typically exhibit wider-than-normal variation are given more latitude) and the typical degree of variation for a given variable.

Registered users can only access a subset of variables from the most recent cost report for each hospital with outliers corrected.

Yes. A documentation zip package is available for download, and contains three Excel workbooks, two .zip files, and a Readme:

• HCRIS_2552-10_Worksheets_A_B_C_D_E_G_L_S_with_varnames_[YYYY_MM_DD].xlsx: CMS hospital cost report worksheets from Form 2552-10, with yellow/green highlights for fields included in processed datasets.
• HCRIS_2552-96_Worksheets__A_B_C_D_E_G_L_S_with_varnames_[YYYY_MM_DD].xlsx: CMS hospital cost report worksheets from Form 2552-96, with yellow/green highlights for fields included in processed datasets.
• P152_36.zip: Chapter 36 of the CMS Provider Reimbursement Manual, which gives hospitals instructions on filling out worksheets in Form 2552-96.
• P152_40.zip: Chapter 40 of the CMS Provider Reimbursement Manual, which gives hospitals instructions on filling out worksheets in Form 2552-10.
• RAND_hospital_data_contents_[YYYY_MM_DD].xlsx: A workbook containing a list of fields included in the datasets available for download for subscribers and registered users.
• Readme.pdf: Contains tips on how to use the included documentation files.