@anon/world-mortality

Dataset

Public

World Mortality

World Mortality Dataset: international data on all-cause mortality.

Sourced from github.com/akarlinsky/world_mortality - contributions are welcomed on the original GitHub repository.

World Mortality Dataset

This repository contains country-level data on all-cause mortality in 2015–2021 collected from the various sources outlined below. Data for 79 countries is currently provided.

Preprint: Karlinsky & Kobak 2021, The World Mortality Dataset: Tracking excess mortality across countries during the COVID-19 pandemic, https://doi.org/10.1101/2021.01.27.21250604

For the excess mortality analysis using this data see github.com/github.com/dkobak/excess-mortality

Notes from the original publishers:

Our aim is to provide data from 2015 onwards. In some cases the coverage starts later, but we require at least full 2019 data.
Countries are only included if the data exist until at least June 2020.
We only collect weekly, monthly, or quarterly data.
The latest data points (weeks/months/quarters) for each country are preliminary and subject to (sometimes large) revisions.
We only provide all-cause mortality numbers, without splitting by age or gender.
We only provide country-level data, without splitting it by regions or individual cities.
The Short Term Mortality Fluctuations (STMF) dataset from the Human Mortality Database (HMD) is integrated into this dataset. See the STMF dataset for mortality by age and gender; here we only provide the total numbers.
The data for the European countries that are not in STMF are sourced from the EuroStat.
Some countries publish obviously incomplete weekly data for most recent weeks, which shows as large "dips" in the end of the time series. We omit these data points for the following countries: Denmark, Finland, South Korea, United States. Note that some other countries may also report partially incomplete data.
Weekly data mostly follow ISO8601 standard, when weeks are calendar weeks, Monday to Sunday, and the weeks on the year boundaries are assigned to the year in which they have more days (four or more). Most years have 52 weeks but some years, such as 2015 and 2020, have 53 weeks. Some countries follow other conventions, see e.g. STMF description.

Sources

Human Mortality Database, Short-Term Mortality Flucations

We collect the weekly STMF data for the following countries: Australia*, Austria, Belgium, Bulgaria, Canada, Chile, Croatia, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Israel, Italy, Latvia, Lithuania, Luxembourg, Netherlands, New Zealand, Norway, Poland, Portugal, South Korea, Russia, Slovakia, Slovenia, Spain, Sweden†, Switzerland, United Kingdom (England & Wales + Northern Ireland†† + Scotland), United States.

We do not use Taiwan data from STMF because the monthly data (see below) is more frequently updated.

United States data from 2017 onwards are replaced with CDC 'predictions' that account for underreporting in recent weeks (see below).

For some European countries, STMF sometimes has more up-to-date (and backward revised) data than Eurostat, as it culls data from countries' NSOs. For each data point that exists in both datasets, we take the maximum between them as the final data.

* Australia's data (all years) is "Doctor Certified Deaths" rather than "All Registered Deaths". These constitute about 85%-90% of all deaths in Australia.

† Sweden has a significant number of deaths which occurred in "unknown" week in all years. 960 deaths in 2015, 1963 in 2016, 2230 in 2017, 2513 in 2018, 2616 in 2019, 2652 in 2020. These are not included in the weekly data.

†† Due to significant reporting delays in Norhtern Ireland data to STMF, we obtained weekly deaths data directly from NISRA. In order to harmonize with England & Wales and Scotland data, data was transformed to ISO-weeks by using the daily average for each week by date start and end.

Eurostat

We collect the weekly data from Eurostat for the following countries: Albania, Armenia, Cyprus, Georgia, Liechtenstein, Malta, Montenegro, Romania, Serbia.

NYT obtained these data directly from the Bolivian officials. Note that the data for April 2020 are missing. To quote NYT, "Bolivia’s Civil Registry recorded almost no deaths in April due to the closure of government offices during a lockdown. Officials said [at] least some of the deaths that occurred in April could have been registered in later months." In their analysis, NYT implicitly assumes that the excess mortality in April was zero.

Bosnia (monthly)

Agency for Statistics of Bosnia & Herzegovina Natural Population Change Quarterly: http://www.bhas.ba/Calendar/Category/14#tab-releases

Brazil (monthly)

2015 to 2017: UNData - http://data.un.org/Data.aspx?d=POP&f=tableCode%3a65%3bcountryCode%3a76%3brefYear%3a2015%2c2016%2c2017&c=2,3,6,8,10,12,13,14&s=_countryEnglishNameOrderBy:asc,refYear:desc,areaCode:asc&v=1
2019 to 2020: Brazilian Population Registry: https://transparencia.registrocivil.org.br/registros

Note: The Brazilian Population Registry has data for monthly mortality in 2018, but it is significantly lower than UNData counts for previous years. This is possibly due to under-reporting in the Registry in early years. We chose not to include 2018 Brazilian currently.

Colombia (weekly)

https://www.dane.gov.co/index.php/estadisticas-por-tema/demografia-y-poblacion/informe-de-seguimiento-defunciones-por-covid-19
Direct link to the latest table in Excel (January 2021): https://www.dane.gov.co/files/investigaciones/poblacion/defunciones-covid19/anexos-defunciones-covid-nal-2020-02mar-2021-17ene.xlsx Here we sum values in three categories: Natural, Violenta (Violent), and En estudio (unclassified deaths).

Costa Rica (monthly)

2015 to 2020: UNData - http://data.un.org/Data.aspx?d=POP&f=tableCode%3a65%3bcountryCode%3a188&c=2,3,6,8,10,12,13,14&s=_countryEnglishNameOrderBy:asc,refYear:desc,areaCode:asc&v=1
2020: Costa Rica National Institute Of Statistics And Censuses (INEC) - Direct link to excel: https://www.inec.cr/sites/default/files/documetos-biblioteca-virtual/repoblacevdef_isem2020.xlsx

The INEC publication provides mortality data for the first half (January-June) of 2020, While UNDATA provides monthly data for January-April. Our estimate for May and June mortality is thus the mean of the difference between the January-June total and the January-June mortality.

Ecuador (weekly)

2017: https://www.ecuadorencifras.gob.ec/nacimientos-y-defunciones-2017/
2018: https://www.ecuadorencifras.gob.ec/category/poblacion-y-demografia/
2019: https://www.ecuadorencifras.gob.ec/defunciones-generales-2019/
2020: https://www.registrocivil.gob.ec/cifrasdefuncion/
Direct link to the latest table in XLS (January 2021):
https://www.registrocivil.gob.ec/wp-content/uploads/downloads/2021/01/Defunciones_Generales_2020_act_24_ENE_2021.xlsx

Ecuador provides daily death counts. We summed them up to form weekly death counts.

Egypt (monthly)

2015 to 2019: