Department of Health & Human Services Hospitalizations
| Attribute | Details |
|---|---|
| Source Name | hhs |
| Data Source | U.S. Department of Health & Human Services |
| Geographic Levels | National, U.S. HHS Region, State (see geography coding docs) |
| Temporal Granularity | Daily (see date format docs) |
| Reporting Cadence | Inactive - no longer updated since 2024-04-30 |
| Date of last data revision: | Never (see data revision docs) |
| Temporal Scope Start | 2019-12-31 |
| License | Public Domain US Government |
Changelog
Click to expand
See COVIDcast Signal Changes for general information about how we track changes to signals.
No changes so far.
Overview
The U.S. Department of Health & Human Services (HHS) publishes several datasets on patient impact and hospital capacity. The data is made available to HHS through the CDC’s National Healthcare Safety Network (NHSN). One of these datasets is mirrored in Epidata at the following endpoint:
- COVID-19 Hospitalization: States - daily resolution, state aggregates
That dataset contains dozens of columns that break down hospital resource usage in different ways.
These indicators make available several commonly-used columns and combinations of columns, aggregated geographically. In particular, we include the sum of all adult and pediatric COVID-19 hospital admissions. This sum is used as the “ground truth” for hospitalizations by the COVID-19 Forecast Hub. We also include influenza hospital admissions.
Each metric below is available in four variants:
- Raw Count:
_1d - Smoothed (7-day average):
_1d_7dav - Population Proportion (per 100k):
_1d_prop - Smoothed Proportion:
_1d_prop_7dav
| Metric | Base Signal Name | Description |
|---|---|---|
| Confirmed COVID-19 | confirmed_admissions_covid |
Sum of Adult + Pediatric. Confirmed admissions only. Earliest Dates: • _1d (Daily): 2019-12-31• _7dav (7-Day Avg): 2020-01-06 |
| Suspected + Confirmed | sum_confirmed_suspected_admissions_covid |
Sum of Adult + Pediatric. Combined count of confirmed and suspected cases. Earliest Dates: • _1d (Daily): 2019-12-31• _7dav (7-Day Avg): 2020-01-06 |
| Influenza | confirmed_admissions_influenza |
All confirmed influenza hospital admissions. Earliest Dates: • _1d (Daily): 2020-01-02• _7dav (7-Day Avg): 2020-01-08 |
Note
- For all the above signals & 7-day average signals, their geography is state, and resolution is 1 day.
- The 7-day average signals are computed by Delphi by calculating moving averages of the preceding 7 days, so e.g. the signal for June 7 is the average of the underlying data for June 1 through 7, inclusive.
Table of contents
Estimation
Statewise, daily resolution, COVID-19
Statewise daily resolution signals for COVID-19 use the following four columns from the HHS state timeseries dataset:
previous_day_admission_[adult|pediatric]_covid_[confirmed|suspected]
The confirmed signal is the sum of the two confirmed columns:
- adult
- pediatric
The sum_confirmed_suspected signal is the sum of all four columns:
- adult confirmed
- adult suspected
- pediatric confirmed
- pediatric suspected
The source data specifies that admissions occurred on the previous day. We automatically adjust the date of each result so that admissions are incident on that date.
Statewise, daily resolution, influenza
Statewise daily resolution signals for influenza use the following column from the HHS state timeseries dataset:
previous_day_admission_influenza_confirmed
The source data specifies that admissions occurred on the previous day. We automatically adjust the date of each result so that admissions are incident on that date.
Limitations
HHS collects data from state and territorial health departments about many, but not all, hospitals in the U.S. Notably excluded from this dataset are psychiatric and rehabilitation facilities, Indian Health Service (IHS) facilities, U.S. Department of Veterans Affairs (VA) facilities, Defense Health Agency (DHA) facilities, and religious non-medical facilities.
Standard errors and sample sizes are not applicable to these metrics.
Lag and Backfill
HHS issues updates to this timeseries once a week, and occasionally more often. We check for updates daily. Lag varies from 0 to 6 days.
Occasionally a value published in an early issue will be changed in a subsequent issue when additional data becomes known. This effect is known as backfill. Backfill is relatively uncommon in this dataset (80% of dates from November 1, 2020 onward are never touched after their first issue) and most such updates occur one to two weeks after information about a date is first published. In rare instances, a value may be updated 10 weeks or more after it is first published.
Source and Licensing
These indicators mirror and lightly aggregate data originally published by the U.S. Department of Health & Human Services. As a work of the US government, the original data is in the public domain.