Data dictionary

Data dictionary: Readmissions files

These files hold what the Readmissions pages show for the FY2027 edition of the Hospital Readmissions Reduction Program (HRRP). hrrp-fy2027-hospitals.csv has one row for each hospital in the program's file: its place, its payment adjustment factor and cut, this site's estimate of the dollars, and, for each of the six conditions, the hospital's cases, its excess readmission ratio and whether it was penalized on the condition. The states file lists the states the same way, and each state, division, region and metro area page has files for the places inside it and for its hospitals. They are files from Paid & Penalized (paidandpenalized.org). They hold no patient information.

schema_version: 1

Source

The payment adjustment factors, cases and excess readmission ratios are CMS's, from its HRRP files for the FY2027 payment year. The Methods page lists every file this site read, with its checksum.
The hospital's name, place and metro area come from this site's hospital master, built from CMS and U.S. Census Bureau files.
The estimated penalty is this site's own estimate, not CMS's number. The Methods page gives every assumption.

Format

UTF-8 without a byte-order mark. Comma-separated (RFC 4180), CRLF line ends, one header row, one row for each hospital (or area). A cell holding a comma, a quote or a line break is quoted.
The CCN is plain text. A spreadsheet may drop its leading zero on opening; import or open the column as text to keep it.
Header names carry their units: percent, dollars, a Y/N flag or a date.
A text cell that begins with =, +, -, @ or a tab gets a leading ', so a spreadsheet reads it as text, not a formula. Number columns are written as they are.
A blank cell means the file has no value; it is never written as 0. A condition's penalized column is blank when the hospital was not measured on the condition: it had fewer than 25 cases of it, or CMS published no ratio for it.

Hospital files

hrrp-fy2027-hospitals.csv, the largest-penalties file on the program page and the area files on the hospital pages share these 31 columns, in this order. An area file (hrrp-fy…-metro-<code>-hospitals.csv for a metro area, hrrp-fy…-state-<xx>-hospitals.csv for the hospitals of a state outside any metro area) lists the hospitals the "other hospitals" table shows on the pages of that area, largest cut first. A place page's hospitals table has the same 31 columns in its own file, largest cut first: hrrp-fy…-place-<kind>-<key>-hospitals.csv, where <kind> is region, division, state or metro and <key> is the place's address (a state by its code, a metro area by its code). A metro area with more than one hospital uses the metro area file above instead, since it holds the same hospitals.

Column	Type	Meaning	Notes
`CCN`	text	The hospital's CMS Certification Number, the ID CMS's files use for it.	The key to join on. Written as plain text, with its leading zero, so open its column as text in a spreadsheet to keep the zero.
`Hospital`	text	The hospital's name, as this site shows it.
`City`	text	The hospital's city.
`State`	text	The hospital's state or territory, as a two-letter code.
`Metro area`	text	The name of the metro area (core-based statistical area) the hospital is in.	Blank for a hospital outside any metro area.
`Beds`	number	The hospital's bed count.	Blank when no source gives one. Certified beds can differ from licensed beds.
`Beds source`	text	Where the bed count comes from.	`certified (Provider of Services file)` or `cost report` (the count from the Medicare cost report, used for the few hospitals without a certified count). Blank when Beds is blank.
`Inactive (date)`	text	The date CMS lists the hospital's Medicare agreement as ended, as YYYY-MM-DD.	`date not listed` when CMS lists the agreement as ended with no date. Blank for a hospital CMS lists as active.
`Payment adjustment factor`	number	The factor CMS applies to the hospital's base payments for the FY2027 payment year.	Four decimals. 1 is no cut.
`Payment reduction (%)`	number	The cut the factor makes, in percent: 100 minus 100 times the factor.	Two decimals. 0 is no cut.
`Estimated penalty ($, modeled)`	number	This site's estimate of the dollars the cut takes, in whole dollars.	An estimate, not CMS's number; the Methods page gives the assumptions. Blank when the hospital has no estimate.
`Peer group`	number	The hospital's peer group, 1 to 5, by the share of its patients who are dually eligible for Medicare and full Medicaid.	Group 1 has the lowest share.
`Dual proportion`	number	The share of the hospital's patients who are dually eligible, as a fraction.	Four decimals. 0.2351 is 23.51%.
`AMI discharges`	number	The hospital's cases for heart attack (acute myocardial infarction) in the FY2027 data window.	Cases are counted whether or not the hospital had enough to be measured.
`AMI ERR`	number	The hospital's excess readmission ratio for heart attack (acute myocardial infarction): its readmissions against what its patients' mix would predict.	Above 1 is more readmissions than predicted. Blank when CMS gave none.
`AMI penalized (Y/N)`	text	`Y` when the hospital's ratio for heart attack (acute myocardial infarction) was above its peer group's median, so the condition counted toward its cut; `N` when it was measured and was not.	Blank, not `0`, when the hospital was not measured on the condition: fewer than 25 cases, or no ratio published by CMS.
`COPD discharges`	number	The hospital's cases for chronic obstructive pulmonary disease in the FY2027 data window.	Cases are counted whether or not the hospital had enough to be measured.
`COPD ERR`	number	The hospital's excess readmission ratio for chronic obstructive pulmonary disease: its readmissions against what its patients' mix would predict.	Above 1 is more readmissions than predicted. Blank when CMS gave none.
`COPD penalized (Y/N)`	text	`Y` when the hospital's ratio for chronic obstructive pulmonary disease was above its peer group's median, so the condition counted toward its cut; `N` when it was measured and was not.	Blank, not `0`, when the hospital was not measured on the condition: fewer than 25 cases, or no ratio published by CMS.
`HF discharges`	number	The hospital's cases for heart failure in the FY2027 data window.	Cases are counted whether or not the hospital had enough to be measured.
`HF ERR`	number	The hospital's excess readmission ratio for heart failure: its readmissions against what its patients' mix would predict.	Above 1 is more readmissions than predicted. Blank when CMS gave none.
`HF penalized (Y/N)`	text	`Y` when the hospital's ratio for heart failure was above its peer group's median, so the condition counted toward its cut; `N` when it was measured and was not.	Blank, not `0`, when the hospital was not measured on the condition: fewer than 25 cases, or no ratio published by CMS.
`PN discharges`	number	The hospital's cases for pneumonia in the FY2027 data window.	Cases are counted whether or not the hospital had enough to be measured.
`PN ERR`	number	The hospital's excess readmission ratio for pneumonia: its readmissions against what its patients' mix would predict.	Above 1 is more readmissions than predicted. Blank when CMS gave none.
`PN penalized (Y/N)`	text	`Y` when the hospital's ratio for pneumonia was above its peer group's median, so the condition counted toward its cut; `N` when it was measured and was not.	Blank, not `0`, when the hospital was not measured on the condition: fewer than 25 cases, or no ratio published by CMS.
`CABG discharges`	number	The hospital's cases for coronary artery bypass graft surgery in the FY2027 data window.	Cases are counted whether or not the hospital had enough to be measured.
`CABG ERR`	number	The hospital's excess readmission ratio for coronary artery bypass graft surgery: its readmissions against what its patients' mix would predict.	Above 1 is more readmissions than predicted. Blank when CMS gave none.
`CABG penalized (Y/N)`	text	`Y` when the hospital's ratio for coronary artery bypass graft surgery was above its peer group's median, so the condition counted toward its cut; `N` when it was measured and was not.	Blank, not `0`, when the hospital was not measured on the condition: fewer than 25 cases, or no ratio published by CMS.
`THA_TKA discharges`	number	The hospital's cases for elective hip or knee replacement in the FY2027 data window.	Cases are counted whether or not the hospital had enough to be measured.
`THA_TKA ERR`	number	The hospital's excess readmission ratio for elective hip or knee replacement: its readmissions against what its patients' mix would predict.	Above 1 is more readmissions than predicted. Blank when CMS gave none.
`THA_TKA penalized (Y/N)`	text	`Y` when the hospital's ratio for elective hip or knee replacement was above its peer group's median, so the condition counted toward its cut; `N` when it was measured and was not.	Blank, not `0`, when the hospital was not measured on the condition: fewer than 25 cases, or no ratio published by CMS.

Area files

The "Every state" file on the program page has these 8 columns, in this order. The files for the places inside a place have the same 8 columns, and differ only in the first column's name: Division (hrrp-fy…-place-region-<key>-divisions.csv, the divisions of a region), State (hrrp-fy…-place-division-<key>-states.csv, the states of a division) or Metro area (hrrp-fy…-place-state-<key>-metros.csv, the metro areas of a state, with a row for the hospitals outside them). A metro area across a state line counts the hospitals in that state only. Every share is over the hospitals measured on at least one condition.

Column	Type	Meaning	Notes
`State`	text	The state's name (the first column's name is `Division` or `Metro area` in the files for those places).	Which hospitals are in a state is as the state's page lists them.
`Hospitals`	number	The hospitals in the area.
`Hospitals measured`	number	Those measured on at least one condition (with 25 or more cases of it and a published ratio).	A hospital measured on nothing cannot take a cut, so every share below is over this count.
`Hospitals penalized`	number	Those with a cut.
`Penalized (% of measured)`	number	Hospitals penalized as a percent of hospitals measured.	One decimal.
`Average cut (%)`	number	The average payment reduction across the area's hospitals, in percent.	Two decimals.
`Cut 1% or more (hospitals)`	number	The hospitals whose cut is 1% or more.
`Estimated penalties ($, modeled)`	number	The sum of the hospitals' estimated penalties, in whole dollars.	Over the hospitals that have an estimate.