Thread by @J_CD_T

THREAD: From today onwards, @TheEconomist will be tracking excess mortality from covid-19 in as many countries as possible. We have published interactive charts for six countries that have released this data. (1/14) economist.com/graphic-detail…

In most countries, the official death toll from covid-19 is a reasonable indicator of whether it is “flattening the curve”. @jburnmurdoch has produced some fantastic charts, tracking these statistics in scores of countries and regions. (2/14)

But official daily death tolls, produced by national health ministries, understate the true number of people killed by the virus at that point. As @OurWorldInData and @MaxCRoser have explained, establishing the cause of death can take several days. (3/14)
ourworldindata.org/coronavirus#re…

Moreover, most official death tolls only include people who tested positive or died in hospitals. As @EasyInve and @webecodibergamo found in Italy, total mortality data show that many other people passed away elsewhere. (4/14)
ecodibergamo.it/stories/bergam…

So we have decided to aggregate data across countries. Using figures from @istat_it, covering 6.7m of 10m people in Lombardy, we found excess mortality of 9,000 in the 4 weeks to March 28th, compared with about 4,000 officially from covid. (5/14)

In Spain, data from @SaludISCIII suggest 13,000 excess deaths in March, compared to 8,000 official. Like Italy, Spain’s daily totals rely chiefly on data provided by hospitals, so might miss many victims who die elsewhere, or were not tested. (6/14)

In France, we used data from @InseeFr, which show excess deaths of about 8,000 people by April 3rd. The official death toll from hospitals was about 5,000 at that point. Since April 1st, the health ministry has included deaths from care homes, causing a large rise. (7/14)

In Britain, data from @ONS suggest excess deaths of about 7,000 in the four weeks to April 3rd. A revised count of covid cases—using death certificates, rather than daily data from hospitals—came to 6,200 in the same period. (8/14)

In the Netherlands, data from @statisticscbs show a large undercount: 4,000 excess deaths in the four weeks to April 5th, vs 1,700 official at that point. This might explain why the current official toll of 3,100 is so low. (9/14)

Most American states are slow to update their total mortality data, but New York City has published weekly data until March 28th via @CDCgov. At this point, they showed an excess mortality of 1,400 in the previous four weeks, vs 1,100 official deaths. (10/14)

However, more recent aggregated data from NYC show that excess mortality in the four weeks to April 11th was at least 7,000, and probably much higher. The city has recently started to count “probable” victims, who died without testing positive. (11/14)
www1.nyc.gov/site/doh/covid…

Most of these data are several days old—unfortunately, the spikes in excess mortality will become much greater in the coming weeks. We will display those as soon as reliable figures come through. (12/14)

This data has taken a while to collect and clean. We hope to publish the code and calculations on Github soon. If you spot any bugs, or any all-cause mortality data for other countries, please email [e-mail] (13/14)

Many thanks to @martgnz for the stunning interactives, to @futuraprime and @MatterofMatt for design tips, and all the academics, researchers and journalists who suggested data. (14/14)