Link Search Menu Expand Document

Hospital Admissions

  • Source name: hospital-admissions
  • First issued: June 30, 2020
  • Number of data revisions since 19 May 2020: 0
  • Date of last change: Never
  • Available for: county, hrr, msa, state (see geography coding docs)

This data source is based on electronic medical records and claims data about hospital admissions, provided to us by health system partners. We use this inpatient data to estimate the percentage of new hospital admissions with a COVID-associated diagnosis code in a given location, on a given day.

Signal Description
smoothed_covid19 Estimated percentage of new hospital admissions with COVID-associated diagnoses, based on data from health system partners, smoothed in time using a Gaussian linear smoother
smoothed_adj_covid19 Same, but with systematic day-of-week effects removed using the same mechanism as in doctor-visits

Table of contents

  1. Limitations
  2. Lag and Backfill
  3. Qualifying Admissions
  4. Estimation
  5. Backfill
  6. Smoothing

Limitations

This data source is based on electronic medical records and claims data provided to us by health system partners. The partners can report on a portion of hospitalizations, but not all of them, and so this source only represents those hospitalizations known to them. Their coverage may vary across the United States.

Lag and Backfill

Hospitalizations are reported and processed by the health system partners several days after they occur, so the signal is typically available within several days of lag. This means that estimates for a specific day are only available several days later.

The amount of lag in reporting can vary, particularly whether the data comes from electronic medical records or from processed claims. After we first report estimates for a specific date, further hospitalization data may arrive for that date, or diagnoses for admissions from that date may change. When this occurs, we issue new estimates. This means that a reported estimate for, say, June 10th may first be available in the API on June 14th and subsequently revised on June 16th.

Qualifying Admissions

We receive the daily count of new hospital admissions recorded by the health system partners at each location. Admissions are considered COVID-associated if they meet the following criteria:

  • If the admission has any ICD-10 code matching {U071, U072, B9729}, or
  • If the primary ICD-10 code is one of {R05, R060, R509, Z9911, R0902, R0603, R0609, R062, R069, R0602, R05, R0600, J9691, J9692, J9621, J9690, J9601, J9600, J189, J22, J1289, J129, J1281, B9721, B9732, B342, B349, A419, R531, R6889} and there is a secondary ICD-10 code of Z20828, or
  • If the primary ICD-10 code is Z20828.

Estimation

For a fixed location and time , let denote the number of hospital admissions meeting the qualifying conditions, and let be the total number of hospital admissions. Our estimate of the COVID-19 percentage is given by

The additional pseudo-observation of 0.5 means this estimate can be interpreted as the posterior mode of a binomial proportion with a Jeffreys prior. The practical effect is to prevent from being exactly zero or one, which would result in estimated standard errors of 0. The estimated standard error is:

Backfill

This source undergoes the same backfill adjustment as the doctor-visits source; see its documentation.

Smoothing

This source undergoes the same smoothing adjustment as the doctor-visits source (see its documentation), with the exception that the smoothing is performed on the raw counts, rather than the raw rate.