Meta Data Name: Access to Hospitals
Date Added: January 6, 2021
Author: Susan Paykin
Date Last Modified: January 3, 2024
Last Modified By: Wataru Morioka
Environment
You can find the variables described in this document in the CSV files here.
CSV files are organized by year and spatial scale. For example, county-level variables from 2000 will be found in C_2000.csv.
Note: Every variable can be found in the Latest files.
Hospital locations were sourced from CovidCareMap Healthcare System Capacity data, an aggregated dataset which sources data from the Healthcare Cost Report Information System (HCRIS) and Definitive Healthcare.
Zip code tract area (ZCTA) and Census Tract files were sourced from the US Census Bureau, TIGER/Line Shapefiles 2018.
The source datatable from CovidCareMap included the hospital name, address, city, state, county, latitude and longitude, as well as additional variables for hospital bed and ICU bed capacity, current bed occupancy rates, and staffing rates. For this analysis, we used only the hospital name and location variables.
Data was cleaned and prepared for analysis. Centroids were calculated for ZCTA and Census Tract geometries. For the nearest resource analysis, Euclidean distance was calculated from the centroid of each tract/ZCTA to the nearest hospital location.
We calculated travel-network access metrics for the driving travel time to the nearest hospital location and count of hospitals within a 30 minute driving threshold. The driving travel cost matrices were sourced from Project OSRM and are available at the Tract or ZCTA scales for multiple transit modes via this Box folder. This analysis was conducted in Python. The script is available in code/Access Metrics - Health Resources.
County and state-level variables include the count of Census tracts and the percent of Census tracts located within a 30 minute driving threshold of hospital, as well as the mean (average) driving time in minutes from Census tracts within the county or state.
- Variable -- title of variable
- Variable ID -- exact name of variable in datasets
- Description -- Short description of variable
- Years Available -- years for which data exists for this variable
- Spatial Scale -- the variable exists for these levels of spatial scale
| Variable | Variable ID in .csv | Description | Years Available | Spatial Scale |
|---|---|---|---|---|
| Distance to nearest hospital | HospMinDis | Euclidean distance* from tract/zip centroid to the nearest hospital, in miles | Latest | Tract, Zip |
| Driving time to nearest hospital | HospTmDr | Driving time from tract/zip origin centroid to the nearest tract/zip hospital destination centroid, in minutes | Latest | Tract, Zip |
| Count of hospitals | HospCntDr | Count of hospitals within a 30-minute driving threshold | Latest | Tract, Zip |
| Variable | Variable ID in .csv | Description | Years Available | Spatial Scale |
|---|---|---|---|---|
| Count of tracts | TotTracts | Total number of tracts in county/state | Latest | County, State |
| Count of tracts within 30-min driving range | HospCtTmDr | Number of tracts with hospital within a 30-min driving range | Latest | County, State |
| Average time drive to nearest hospital | HospAvTmDr | Average driving time (minutes) across tracts in county/state to nearest hospital | Latest | County, State |
| Percent of tracts within 30-min driving range | HospTmDrP | Percent of tracts with hospital within a 30-min driving range | Latest | County, State |
*Euclidean distance or straight-line distance is a simple approximation of distance or travel time from an origin centroid to the nearest health center. It is not a precise calculation of real travel times or distances.
This dataset includes all US states, Washington D.C., and Puerto Rico. It does not include the territories of Guam, Northern Mariana Islands, American Samoa, and Palau. Zip code and tract centroids are not population-weighted.