NCHS Mortality Data

This data source of national provisional death counts is based on death certificate data received and coded by the National Center for Health Statistics (NCHS). This data is different from the death data available from USAFacts and JHU CSSE: deaths are reported by the date they occur, not the date they are reported by local health departments, and data is frequently reissued as additional death certificates from recent weeks are received and tabulated.

Signal Description
deaths_covid_incidence_num Number of weekly new deaths with confirmed or presumed COVID-19
Earliest date available: Epiweek 05 2020
deaths_covid_incidence_prop Number of weekly new deaths with confirmed or presumed COVID-19, per 100,000 population
Earliest date available: Epiweek 05 2020
deaths_allcause_incidence_num Number of weekly new deaths from all causes
Earliest date available: Epiweek 05 2020
deaths_allcause_incidence_prop Number of weekly new deaths from all causes, per 100,000 population
Earliest date available: Epiweek 05 2020
deaths_flu_incidence_num Number of weekly new deaths involving influenza, including those with COVID-19 or pneumonia also listed as a cause of death
Earliest date available: Epiweek 05 2020
deaths_flu_incidence_prop Number of weekly new deaths involving influenza, including those with COVID-19 or pneumonia also listed as a cause of death, per 100,000 population
Earliest date available: Epiweek 05 2020
deaths_pneumonia_notflu_incidence_num Number of weekly new deaths involving Pneumonia, excluding Influenza deaths
Earliest date available: Epiweek 05 2020
deaths_pneumonia_notflu_incidence_prop Number of weekly new deaths involving Pneumonia, excluding Influenza deaths, per 100,000 population
Earliest date available: Epiweek 05 2020
deaths_covid_and_pneumonia_notflu_incidence_num Number of weekly new deaths involving COVID-19 and Pneumonia, excluding Influenza
Earliest date available: Epiweek 05 2020
deaths_covid_and_pneumonia_notflu_incidence_prop Number of weekly new deaths involving COVID-19 and Pneumonia, excluding Influenza, per 100,000 population
Earliest date available: Epiweek 05 2020
deaths_pneumonia_or_flu_or_covid_incidence_num Number of weekly new deaths involving Pneumonia, Influenza, or COVID-19
Earliest date available: Epiweek 05 2020
deaths_pneumonia_or_flu_or_covid_incidence_prop Number of weekly new deaths involving Pneumonia, Influenza, or COVID-19, per 100,000 population
Earliest date available: Epiweek 05 2020
deaths_percent_of_expected Number of weekly new deaths for all causes in 2020 compared to the average number across the same week in 2017–2019
Earliest date available: Epiweek 05 2020

Table of contents

  1. Calculation
  2. Geographical Exceptions
  3. Report Using Epiweeks
  4. Missingness
  5. Lag and Backfill
  6. Limitations
  7. Source and Licensing

Calculation

These signals are taken directly from Table 1 without changes. National provisional death counts include deaths occurring within the 50 states and the District of Columbia that have been received and coded as of the date specified during a given time period. The deaths are classified based on a new ICD-10 code. (Note that the classification is based on all the codes on the death certificate, not just the primary cause of death). The codes that are considered for each signals are described in detail here. We export the state-level data as-is in a weekly format.

Geographical Exceptions

New York City is listed as its own region in the NCHS Mortality data, but we don’t consider NYC separately. The death counts for NYC are included in New York State in our reports.

Report Using Epiweeks

We report the NCHS Mortality data in a weekly format (time_type=week & time_value={YYYYWW}, where YYYYWW refers to an epiweek). The CDC defines the epiweek as seven days, from Sunday to Saturday. We check the week-ending dates provided in the NCHS morality data and use Python package epiweeks to convert them into epiweek format.

Missingness

NCHS suppresses some data to protect individual privacy and avoid publishing low-confidence figures. This includes data for jurisdictions where counts are between 1 and 9, and data for weeks where the counts are less than 50% of the expected number, since these provisional counts are highly incomplete and potentially misleading.

Lag and Backfill

There is a lag in time between when the death occurred and when the death certificate is completed, submitted to NCHS, and processed for reporting purposes. The death counts for recent weeks are continually revised and may increase or decrease as new and updated death certificate data are received from the states by NCHS. This delay can range from 1 to 8 weeks or even more; review this detailed analysis of reporting delays for more information. Some states report deaths on a daily basis, while other states report deaths weekly or monthly. State vital record reporting may also be affected or delayed by COVID-19 related response activities which make death counts not comparable across states. We check for updates reported by NCHS every weekday but will report the signals weekly (on Monday).

Limitations

Standard errors and sample sizes are not applicable to these metrics.

Source and Licensing

This data was originally published by the National Center for Health Statistics, and is made available here as a convenience to the forecasting community under the terms of the original license. The NCHS places restrictions on how this dataset may be used: you may not attempt to identify any individual included in the data, whether by itself or through linking to other individually identifiable data; you may only use the dataset for statistical reporting and analysis. The full text of the NCHS Data Use Agreement is available from their website.