If Oklahoma's weakest rural hospitals closed: drive times and beds, a scenario
Copyright 2026 Craig Muldoon. Made with Plotlay (https://www.getplotlay.com).

This repository contains code, data we made, and data from the sources below.

CODE
  The notebook and the Power BI project (its Power Query, DAX and report files) are licensed under the MIT licence
  (LICENSE).

DATA WE MADE
  The tables in data/tables/ (the risk rule's inputs and results, the closures history, each block group's drive times,
  each county's roll-up, the scenario's beds), the shapes in data/geometry/ and the visuals are licensed under Creative
  Commons Attribution 4.0 International (LICENSE-DATA). Please credit: "Rural hospital closures in Oklahoma: Plotlay
  (getplotlay.com), from CMS, HHS and U.S. Census Bureau data". They are made from the sources below.

UPSTREAM SOURCES
  All are works of the United States Government, public domain in the United States (17 U.S.C. 105).

  Centers for Medicare & Medicaid Services: Hospital Provider Cost Report (the HCRIS public use file), fiscal years
  2019 to 2024, data.cms.gov.
  https://data.cms.gov/provider-compliance/cost-reports/hospital-provider-cost-report
  Changes: total margin (Net Income / (Net Patient Revenue + Total Other Income)), patient margin and Medicaid share
  of charges worked out by us; a Medicaid share above 100% treated as a reporting error (blank).

  Centers for Medicare & Medicaid Services: Provider of Services File, Hospital and Non-Hospital Facilities
  (Quality Improvement and Evaluation System), third quarter 2026, data.cms.gov.
  https://data.cms.gov/provider-characteristics/hospitals-and-other-facilities/provider-of-services-file-quality-improvement-and-evaluation-system
  Changes: rural short-term and critical access hospitals whose Medicare agreement ended since 1 January 2005,
  classed by us by what the file shows at the same address or town afterwards (continued, closed, closed and later
  reopened, became a Rural Emergency Hospital).

  U.S. Department of Health and Human Services: COVID-19 Reported Patient Impact and Hospital Capacity by Facility,
  dataset anag-cw7u, healthdata.gov, weeks of 30 April 2023 to 27 April 2024 (the last year of reporting), for
  Oklahoma, Texas, Kansas, Arkansas, Missouri, Colorado and New Mexico.
  https://healthdata.gov/Hospital/COVID-19-Reported-Patient-Impact-and-Hospital-Capa/anag-cw7u
  Changes: each hospital's mean inpatient beds and beds in use over the weeks it reported; suppressed small counts
  (-999999) left out, never read as zero; its geocoded address used as its place.

  U.S. Census Bureau: American Community Survey 2020-2024 5-year estimates, table B01003 (total population), block
  groups; 2020 Census centers of population by block group (CenPop2020); cartographic boundary files 2024 (block
  groups, counties, states, 1:500,000); TIGER/Line 2024 roads; the 2024 ZCTA gazetteer; the Census Geocoder.
  Changes: drive times worked out by us with Plotlay's drive-time engine over the roads (speeds by road class stated
  in the README); outlines simplified.

  U.S. Department of Agriculture, Economic Research Service: 2020 Rural-Urban Commuting Area Codes (ZIP codes and
  census tracts). Used to tell rural from urban.

  Public Law 119-21 (H.R. 1, approved 4 July 2025), text from the U.S. Government Publishing Office (govinfo.gov),
  quoted in the README.

  Base map in the Plotlay project and the app screenshots
    USGS The National Map: US Topo (public domain).

CITED, NOT REPUBLISHED
  The Cecil G. Sheps Center for Health Services Research, University of North Carolina at Chapel Hill, keeps a list of
  rural hospital closures (https://www.shepscenter.unc.edu/programs-projects/rural-health/rural-hospital-closures/).
  Its data is not licensed for republishing, so none of it is in this repository: the README compares our counts with
  its published totals only.

NOT INCLUDED
  No patient-level data: every value is per hospital, per block group or per county. Patients moved in the scenario
  are average patients a day, not people. The live drive-time run near McAlester in the Plotlay project used
  OpenStreetMap roads (ODbL); nothing in data/ is derived from OpenStreetMap. No imagery was traced, and no Google,
  Nearmap, Esri or Mapbox imagery was used. Nothing here implies endorsement by CMS, HHS, the Census Bureau, the
  Sheps Center or any hospital.

THIRD-PARTY SOFTWARE
  The notebook uses pandas, GeoPandas, Shapely, pyproj and Matplotlib (installed from requirements.txt, each under its
  own licence). Fonts in the visuals: IBM Plex Sans and IBM Plex Sans Condensed, SIL Open Font License 1.1.
