HIFLD/Public Health

From OpenStreetMap Wiki
Jump to navigation Jump to search

The purpose of this page is to document the import of the data in the Public Health category of the HIFLD.

Suitable Data

Name Quality Notes Status Size
Hospitals Planned 8,013

License

Due to the ambiguous phrasing of the license statement:

None (Public Use). Users are advised to read the data set's metadata thoroughly to understand appropriate use and data limitations.

I clarified this in an email to HIFLD@hq.dhs.gov. Their response can be found below:

Thank you for your inquiry, William. The dataset you are referring to is being provided for public use.

Best regards,

HIFLD Support Team

Geospatial Management Office (GMO) | Office of the Chief Data Officer (OCDO)

Office of the Chief Information Officer (OCIO) |Department of Homeland Security (DHS)

Hospitals

Tagging

HIFLD Tag OSM Tag Notes
NAME name
ADDRESS addr:* This is parsed into addr:housenumber and addr:street using the usaddress library,

but also provides the original field under addr:full for reference. The library uses statistical methods and isn't 100% accurate, so this field should be manually checked to ensure the parsing worked as expected. Some addresses can't be parsed by the library used. addr:full should then be deleted from the feature after it is verified.

CITY addr:city
STATE addr:state I usually remove this from the field because it's redundant with state admin boundaries.
ZIP addr:postcode
ZIP4 addr:postcode This is appended to addr:postcode with a hyphen when provided.
TELEPHONE phone US country code +1 is added and the phone number is converted to use space separators instead of hyphens and parentheses.
WEBSITE website Trailing slashes are removed and https:// is preprended if an http(s) prefix isn't already present.

On conflict, I check both website URLs to see which one is accurate.

STATUS amenity

disused:amenity

For closed hospitals, disused:amenity=hospital is used. Otherwise, amenity=hospital is used.

When a conflict arises on this field, I check the website status to see if the hospital is really closed or not. If the hospital has been converted into a clinic (i.e. does not provide inpatient care), the disused tag is still appropriate.

BEDS beds On conflict, check the website to see if this data is available there. If not, use the existing tagged number.

I've found a few places where the existing tag is accurate and the HIFLD data is not.

TYPE healthcare=rehabilitation

healthcare=psychiatry

Currently in the script:
  • TYPE = PSYCHIATRIC -> healthcare=psychiatry
  • TYPE = REHABILITATION -> healthcare=rehabilitation

Not handled in the script - needs further discussion:

  • TYPE = CHILDREN -> ??? (I think healthcare:speciality=paediatrics would be appropriate, but it's not in the script yet)
  • TYPE = GENERAL ACUTE CARE ->???
  • TYPE = CRITICAL ACCESS ->???
  • TYPE = LONG TERM CARE ->???
  • TYPE = MILITARY ->???
  • TYPE = SPECIAL ->???
  • TYPE = WOMEN ->???
  • TYPE = SPECIAL ->???
TRAUMA Does being a certified trauma center imply emergency=yes? Currently this is not handled in the script, but should be considered.

Is there a trauma center certification tag in OSM? There's also a healthcare=trauma tag, but it's meant specifically for trauma surgery, and not to indicate a trauma center designation.

Quality

TODO: do an objective measure of quality on 100 data points.

So far the quality of these fields seems to be fairly good, but some areas need manual review before importing. It's important to consider whether entries should be tagged as a hospital or a clinic. The main distinction is that hospitals offer inpatient care. Some entries from this dataset are actually clinics, or were shut down as a hospital and now serve as clinics. This can also be inferred based on the size of the building, type of building (a hospital usually isn't in a retail space), but the website listing provided services is the best way to differentiate. The hours listed can also be helpful for this.

The website field requires special consideration, because hospitals frequently get bought by other organizations or merged, and the website changes domains. Sometimes a search engine is needed to find the new website for a hospital.

The addresses seem especially accurate. There were very few places where address fields were conflicting with existing tags.