Apr 16th 2020
AS COVID-19 has spread around the world, people have become grimly familiar with the death tolls that their governments publish each day. These numbers give a better indication of a country’s trajectory than do counts of confirmed cases, which largely measure how many people have been tested. Nonetheless, official covid-19 death tolls still under-count the true number of fatalities that the disease has already caused at that point.
In many places, official daily figures exclude anybody who did not die in hospital or who did not test positive. Often the cause of death takes several days to establish, which creates a lag in the data. And even the most complete covid-19 records will not count people who were killed by other conditions that probably would have been treated successfully, had hospitals not been overwhelmed by a surge of covid-19 cases.
The best way to measure the full damage caused by such a medical crisis is to look at “excess mortality”: the gap between the total number of people who died from any cause during a given period, and the historical average for the same place and time of year. These numbers take longer to emerge. Some countries publish them with a few days’ delay, others after more than a year. EuroMOMO, a group of academics from 24 European countries, releases a weekly index. But it does not publish absolute numbers of deaths for each country.
However, the national statistical offices in a handful of Western countries have released some total mortality data to the end of March, as we first reported two weeks ago. A few—such as Italy and the Netherlands—showed an excess of deaths that was more than twice the official covid-19 tally for the same period. That discrepancy will surely be greater in poorer countries, which have less capacity for testing and treating patients. As more places start to publish their total mortality figures, The Economist will report and analyse them on this page.
Italy was the first European country to be overrun by the virus. On April 15th its official count of covid-19 deaths was over 21,000. Yet the true toll might have been double that. Giorgio Gori, the mayor of Bergamo, a northern town hit especially hard, has said that the official data are “the tip of the iceberg…too many victims are not included in the reports because they die at home.” The official covid-19 tally, published by the ministry of health, includes only people who tested positive before passing away.
So far, Italy’s national statistical bureau has published figures of deaths from all causes for only 1,450 of the country’s 7,900 municipalities, covering the period until March 28th. We have analysed a sample of these areas, including Bergamo, that contain 6.7m of the 10m residents in Lombardy, a region that has suffered nearly half the country’s official fatalities. By March 28th, the excess deaths in this sample had reached more than 9,000 in the previous five weeks, covering the entire period of the outbreak. At that point, the official covid-19 tally was 4,000. This suggests that the true toll was about 120% higher.
Spain has experienced a similar epidemic, with its official death toll approaching 19,000 on April 15th, according to data from hospitals. Like Italy, the country’s overall mortality data imply that the actual number is higher, although the extent of the undercounting is less severe. Spain’s national epidemiology centre is publishing regular figures for deaths from all causes in each region. By March 31st, these showed that there had been 13,000 excess fatalities nationwide that month, covering the entire period of the outbreak. The official covid-19 tally at that point was 8,000. This suggests that the true toll was about 60% higher.
France’s official death toll accelerated rapidly in April, after its central health authority began including people who died in nursing and care homes. By April 15th, nearly 40% of the 17,000 recorded national fatalities came from such institutions. This approach should mean that from now on the country will understate covid-19 deaths less severely than its European peers.
By April 3rd, the latest date for which the national statistical bureau has published data for deaths from all causes, there had been an excess of nearly 8,000 fatalities during the previous four weeks, covering the entire period of the outbreak. The official covid-19 tally from hospitals alone at that point was 5,000. This suggests that the true toll was about 60% higher.
Britain’s national statistical office is publishing data for deaths from all causes. It has also started to produce retrospective estimates of how many people died of covid-19 on any given day, after analysing the probable causes mentioned on death certificates. These revised covid numbers are shown in our charts. They are much higher than those published daily by the health ministry, which rely on the latest submissions from hospitals (and reached nearly 13,000 by April 15th).
By April 3rd, the latest date for which deaths from all causes are available, there had been around 7,000 excess deaths in England and Wales during the previous four weeks, covering the entire period of the outbreak. The revised covid death toll was 6,200—up from the 4,300 recorded in the health ministry’s daily data. This suggests that the true death toll was still about 10% higher than the revised covid tally.
So far, the number of official covid-19 deaths in the Netherlands has shown a much flatter trajectory than in other western European countries. On April 15th the tally stood at 3,100. However, this is because the national institute for public health, which publishes the daily figures, includes only people who have tested positive for the virus and died in hospital (often with a delay). By April 5th, the latest date for which the national statistical bureau has published data about deaths from all causes, there had been 4,000 excess deaths during the previous four weeks, covering the entire period of the outbreak. The official covid-19 tally at that point was 1,700. This suggests that the true toll was about 140% higher.
After trailing behind the death tolls of many European countries at first, America’s official death count has risen sharply. By April 15th it had recorded 28,000 victims, more than any other country. Though most American states do not publish recent records of deaths from all causes, such data are produced by the city of New York, the worst-affected area. Our chart shows its latest weekly figures, which ended on March 28th. These demonstrate an excess of about 1,400 deaths during the previous four weeks, covering the entire period of the outbreak, compared with 1,100 official covid-19 fatalities at that point.
The city has published some more recent data on total mortality, aggregated over a longer period, which suggest that excess mortality in the four weeks to April 11th was at least 7,000 (and probably much higher). We will show those data in our charts once they have been finalised and broken down by week.
On April 14th New York’s health department adopted a similar approach to Britain’s statisticians, by expanding its definition of covid deaths and analysing death certificates retrospectively. It found an additional 4,000 people who had “probably” died of the virus, to go with the 7,000 victims who had tested positive at that point. This should make the official covid-19 figures resemble the excess deaths more closely from now on.
Sources: ECDC; ISTAT; Ministero della Salute; Instituto de Salud Carlos III; Datadista; INSEE; Santé Publique France; ONS; Centraal Bureau van Statistiek; CDC; New York City Health