An example Stata extract code is also available. Some materials are set up as well.
- HCRIS Skilled Nursing Facility (SNF)
- HCRIS Renal Dialysis Facility
- HCRIS Hospice
- HCRIS Home Health Agency (HHA)
- HCRIS Health Clinic
- HCRIS Community Mental Health Center
- HCRIS Freestanding Federally Qualified Health Center
CMS provides a 2552-96 to 2552-10 crosswalk and a list of cost-coded worksheets. The text files in 2552-10 and 2552-96 may also be helpful in getting an idea of what lines are on each version of the worksheets. The text files in 2552-10 and 2552-96 should be helpful for getting an idea of what worksheet, column, and line combinations are on each version of the worksheets. Per the README, the 2552-96 lines are changed to the Cost Center Codes for the cost center coded worksheets. In 2552-10, the cost-coded line numbers are not changed to cost codes. Instead, the cost center associated with each cost-coded line number is described in the alphanumeric data file value for the report number where WKSHT_CD = "A000000" and CLMN_NUM = "00000". Another difference between 2552-96 and 2552-10 is that the column number variable was 4 columns wide in 2552-96 and 5 columns wide in 2552-10. So, for example, column 1 in 2552-96 is "0100" and "00100" in 2552-10.
CMS began including rollups of their internal 2552-10 A, B, C, D, E, G, and S series in April 2014. They consider the 2552-96 rollup files to be unreliable.
CMS released 2552-96 bad debt files but not 2552-10. However, a 2552-10 bad debt bad debt rosetta stone is available.
Select variables from the cost reports is available in the files below. One of the variables included is ZIP Code. A ZIP Code distance database is available.
|Cost Report||Hospital Provider and Address Info||Hospital Cost Report Status Counts||Hospital Record Counts||Worksheet Column Line Combinations|
Complete Data Files in a Long Shape
Select Variables in a Flat Shape
To get the worksheets that were completed to generate the cost report data, go to Paper-Based Manuals, choose Publication # 15-2 for the Provider
To get the worksheets that were completed to generate the cost report data, go to Paper-Based Manuals, choose Publication # 15-2 for the Provider Reimbursement Manual Part 2, then choose Chapter 36. R20P236F.zip in P152_36.zip has the 2552-96 worksheets A-M,S for "cost reporting periods ending on or after September 30, 1996" or Chapter 40 for "cost reports with fiscal years beginning on or after May 1, 2010"..
The layout for the way 2552-96 cost report data was delivered might be helpful for getting an idea of some of the variables in some of the worksheets.
The HCRIS data consists of four databases: one has alphanumeric variables, one has numeric variables, one has hospital report meta-variables, and one has many of the individual numeric variables "rolled up" into one variable.
Be careful when using both the rollup file or the numeric file with negative amounts. For some items, losses are to be recorded as (+amount). In the past, sometimes values get recorded as -amount instead. Check worksheet instructions and the data when there may be negative amounts.
The primary key linking these datasets is the report record number, RPT_REC_NUM. The hospital report database is an ordinary and small rectangular data file. The alphanumeric (A) and especially the numeric (N) databases are big, long, skinny files. The A & N files have all the HCRIS report variables for all fiscal years from 1996 on. They have five variables each: RPT_REC_NUM, WKSHT_CD, LINE_NUM, CLMN_NUM, and the value. Extract data from these files using the worksheet code, line number and column number. Worksheets have names like 255296_*.xls, where * is a letter, a through s. The rollup files have three variables: RPT_REC_NUM, LABEL which is a reasonable mnemonic variable name reflecting the worksheet code, line number, and column numbers that were rolled up, and ITEM which is the value.
A statistical or database package such SAS or Oracle, etc. , or a programming language that can handle large files is necessary to use the HCRIS data. The fyYEAR.zip files are about 100 Mb and can unzip to over 1 Gb.
The SAS datasets created by these programs can be converted to other formats using conversion software such as Stat/Transfer.
- Worksheet Name / Worksheet Code Crosswalk
- Rolled Up Variables and Variable Extraction
- SAS Programs
- Comparing HCRIS variable combinations to PPS
The file below provides links between the seven-character worksheet codes, WKSHT_CD, used in the alphanumeric and numeric databases and descriptive labels. Over 1,300 codes are listed. For example,
- A83P001 A-8-3, Part I Physical Therapy (h)
|wk_xwalk.txt||Crosswalk between worksheet name and worksheet code, WKSHT_CD|
The files below may be helpful for extracting data. They list every combination of worksheet code, column number, and line number in HCRIS' alphanumeric and numeric databases. These files can be manipulated using software such as MS Excel ( Especially the alphanumeric file. MS Excel has a limit of 65536 lines and the numeric file has over 200,000 unique combinations of WKSHT_CD, CLMN_NUM, and LINE_NUM. The numeric file could be cut down using an editor, or read with a package such as MS Access. )
SAS-friendly versions are already available with the SAS programs below.
|alphnmrc||Alphanumeric worksheet codes, column numbers, and line number combinations|
|nmrc||Numeric worksheet codes, column numbers, and line number combinations|
(Note to NBER internal users: The first four programs have already been run. Begin by modifying the extract files to include specific variables of interest).
|read_hosp_rpt.sas||read HCRIS' hosp_rpt.zip and create hosp_rpt.sas7bdat|
|read_hosp_rpt_alphnmrc.sas||read HCRIS' hosp_YEAR_ALPHNMRC.CSV files and create hosp_alpha_rptYEAR_long.sas7bdat files|
|read_hosp_rpt_nmrc.sas||read HCRIS' hosp_YEAR_NMRC.CSV files and create hosp_nmrc_rptYEAR_long.sas7bdat files|
|read_hosp_rpt_rollup.sas||read HCRIS' hosp_YEAR_ROLLUP files and create hosp_rollupYEAR.sas7bdat files|
|extract sample alphanumeric variables from hosp_alpha_long.sas7bdat to alpha.sas7bdat|
|alpha where||SAS "WHERE" statement for all alphanumeric WKSHT_CD, CLMN_NUM, LINE_NUM combinations ( cut and paste needed lines into WHERE statement of alpha.sas )|
|alpha when||SAS "WHEN" statement for all alphanumeric WKSHT_CD, CLMN_NUM, LINE_NUM combinations ( cut and paste needed lines into WHEN statement of alpha.sas )|
|extract sample numeric variables from hosp_nmrc_long.sas7bdat to nmrc.sas7bdat|
|numeric where||SAS "WHERE" statement for all WKSHT_CD, CLMN_NUM, LINE_NUM combinations ( 16 Mb ) ( cut and paste needed lines into WHERE statement of nmrc.sas )|
|numeric when||SAS "WHEN" statement for all numeric WKSHT_CD, CLMN_NUM, LINE_NUM combinations ( 25 Mb ) ( cut and paste needed lines into WHEN statement of nmrc.sas )|
|transpose_alpha.sas||transpose to wide shape (multiple variables per report number), alphat.sas7bdat|
|transpose_nmrc.sas||transpose to wide shape (multiple variables per report number), nmrct.sas7bdat|
|merge.sas||merge alphanumeric and numeric datasets to hospital report datasets by fiscal year|
Users of the old Prospective Payment System PPS version of the hospital cost report data may wonder how PPS fields correspond to data extracted from HCRIS. The worksheet codes, column numbers, and line numbers from the most recent ( fiscal years 1996-1999 ) PPS files correspond exactly to the earliest 1996-1999 HCRIS files. Most PPS fields correspond to _exactly__one_ column and line on a 2552-96 worksheet. A few PPS fields, however, are the sum of multiple lines. The means files below are one way to check that HCRIS-extracted data corresponds to PPS fields. The match may not be exact, though. The HCRIS files have the most up-to-date cost report data for fiscal years 1996-1999. Frequecy tables of the character variables are included as well.
|PPS Fiscal Year||Medians of Numeric Variables||Means of Numeric Variables||Frequencies of Character Variables|
Send questions to firstname.lastname@example.org.