Showcase · Data layers · HHS and U.S. Census open data

How full were LA County’s hospitals?

How full were Los Angeles County’s hospitals through the pandemic and after, and how far do people live from one? Every hospital that reported to the U.S. Department of Health and Human Services, week by week from June 2020 to April 2024, and every census tract by its distance to a general hospital.

In the week of 3 January 2021 the county’s hospitals held 8,316 COVID-19 patients. 80.6% of their 19,773 inpatient beds were in use, and 32 of the 83 hospitals reporting were 90% full or more. The surge passed; the strain stayed in some places. In 2023 and 2024, 16 of 74 general hospitals averaged 90% full or more.

Hospitals in rows, people in tracts

HHS published every hospital each week as a row: its beds, the beds in use, its COVID-19 patients and its staffed adult ICU beds, with an address. A row has no area and no people. The Census Bureau counts people by tract, and its tracts hold no hospitals. Answering “how full, and for whom?” needs both on one map. Here the hospitals are keyed by their CMS number and the tracts by their GEOID, and the week is on a slider.

The map

Each hospital is a dot, sized by its beds and coloured by the share of its inpatient beds in use in the week on the slider. Under them each census tract is shaded by its straight-line distance to the nearest general hospital, or by the beds within 10 km for every 1,000 people. Press Play to step through 203 weeks. Hover over a hospital or a tract, or tap it on a phone, for its numbers.

The surge, and the beds after it

COVID-19 patients in the county’s hospitals rose from 1,950 in the week of 7 June 2020 to 8,316 in the week of 3 January 2021. Beds in use peaked a week later at 16,252 (83.4%), and staffed adult ICU beds were 91.6% in use in the week of 17 January.

Two years later the share in use was back where it had been at the peak. In the week of 5 March 2023, with 609 COVID-19 patients, 80.9% of the county’s beds were in use. Fewer patients filled a smaller count of beds: the hospitals reported 17,601 inpatient beds that week, against 19,773 in the week of 3 January 2021, when surge beds were open. The data can’t say what filled the beds; it counts beds and patients, not why people were in hospital.

The fullest hospitals after the pandemic

Across all weeks of 2023 and 2024 the county’s beds were 77.1% in use in the median week. The average hides the hospitals that ran full: of the general hospitals with 100 beds or more, Martin Luther King, Jr. Community Hospital (98.1%), Ronald Reagan UCLA Medical Center (97.5%), USC Arcadia Hospital (96.8%), City of Hope Helford Clinical Research Hospital (96.2%). City of Hope’s Duarte hospital, a cancer centre, was 90% full or more in every one of the 203 weeks.

How far people live from a general hospital

Half of the county’s 9,936,690 residents live within 2.84 km (1.8 miles) of a general hospital in a straight line, and 82.9% within 5 km. 258,746 people live 10 km or more away, most of them in the Antelope Valley north of the mountains, where the tracts nearest Antelope Valley Hospital or Palmdale Regional Medical Center hold 411,291 people. There each resident has 1.09 beds within 10 km for every 1,000 people, against 1.67 for the county. Antelope Valley Hospital was 90% full or more in 190 of 203 weeks, and 95.4% full on average in 2023-24.

These are straight lines. A drive is longer, and over the San Gabriel Mountains or in freeway traffic much longer. Walking and drive-time catchments are not in Plotlay yet; when they are, this map will use them.

How it was made in Plotlay

1. A map project on the USGS National Map topo base map (public domain), and two data layers.
2. Hospitals: Take all. One file with each of the 86 hospitals as a disc 1 km across at its HHS address, big enough to see across the county: all of them in one step, keyed by their CMS number.
3. Link by key each hospital’s summary, then 17,031 rows of hospital weeks as a time axis, each of the 203 weeks a stop.
4. A formula column: the weeks each hospital was 90% full or more, count(occupancy >= 90).
5. Colour by the share of beds in use for the week on the slider, and play it.
6. Tracts: Take all from the Census file: 2,495 tracts keyed by GEOID, linked to their people and their distance to a hospital, coloured by distance.
7. Charts and map pictures in the Analysis workspace, exported as PNG; both layers exported as GeoJSON, TopoJSON and CSV with WKT.

More about Data layers · Make a data layer of your own

Power BI

The same tables as a Power BI Project (.pbip): the county week by week, one week hospital by hospital on Azure Maps with a week slicer, the hospitals after the pandemic, and a Shape Map of the tracts by distance, with the Plotlay theme. It reads the data from the GitHub repository, so it refreshes with no gateway. It is in the plotlay-la-hospital-beds repository on GitHub, with the steps to open it.

Python

A narrated Jupyter notebook reads the tables and shapes, recomputes every number on this page, and maps the tracts and the hospitals.

Download the notebook (with its outputs) and the data. Both are also in the repository on GitHub.

Read it carefully

• The data stopped in May 2024, when hospitals no longer had to report to HHS. Its last week is 21 April 2024. A week is a 7-day average, Sunday to Saturday.
• Beds are what each hospital reported: staffed inpatient beds, surge and overflow beds included. The count fell after the surge, so a share in use can hold steady while fewer patients are in hospital.
• The bed counts step down in the week of 12 July 2020, the first week the file reports staffed ICU beds, so the beds chart starts there.
• Small counts are suppressed. HHS hides counts under 4; they are blank here, never zero. County totals add up the hospitals that reported that week (about 84).
• Two records are not hospitals of their own. One repeated Casa Colina Hospital week for week from December 2022, so it is dropped; a 42-bed temporary site at Long Beach Medical Center in late 2020 is added to that hospital.
• Distances are straight lines from a point inside each tract to the nearest general hospital (short term or critical access) reporting in 2023-24, not drive times. Beds per 1,000 people within 10 km shares each hospital’s beds among everyone within 10 km of it; it ignores the county line, so the edges of the county read low.
• No patient-level data. Every value is a weekly total per hospital as HHS published it. Names are HHS’s, with acronyms restored (LAC, UCLA, USC, PIH, UCI).

The other United States showcases

• Which buildings face Local Law 97 fines?: every benchmarked New York City building’s emissions against its limit.
• How far has the soft-story program got?: San Francisco’s retrofits, parcel by parcel.