Dying FOR the coronavirus or WITH the coronavirus?

The real number of deaths in Italy

In Bergamo, people are dying at 4.5 times the rate of a normal year. In the 1st-24th of March period, 446 deaths in 2020 vs 98 in the same period in 2019.

(source and more data below)

We recently learned that more of 85% of COVID-19 victims in Italy had one or more pre-existing health conditions (accounts differ on the precise percentage).

Does it mean that we are overestimating how many people die because of the coronavirus?

I suspect the opposite: we are underestimating.

For example, the mayor of Bergamo said: "For each COVID death there are 3 who die at home of pneumonia without a test" (link).

How can we know how many people are really dying because of the COVID-19?

The easiest way to know, and perhaps the best one, is to compare March deaths in 2020 vs March deaths in 2019 over the same geographical area.

This is the only measure which would take into account both direct deaths (they died because of complications caused by the virus) and indirect ones (they died because of unrelated health conditions but would have not died if not for the pandemic which is overloading the healthcare system and decreasing the amount of healthcare attention that citizens can get).

Unfortunately, I couldn't find any good source for the whole of Lombardy. However, these articles (link, link, link and link) provided data given by mayors and other officials of Bergamo and of the towns around it. Let's see what they say.

In Bergamo, people are dying at 4.5 times the rate of a normal year. In the 1st-24th of March period, 446 deaths in 2020 vs 98 in the same period in 2019 (link).

70 people died in about the first 3 weeks of March in Dalmine vs 18 last year.

Of course, March 2020 might have been a bad flu year compared to March 2019 even if the coronavirus outbreak didn't take place. However, 175-215 people die each year in Dalmine (link). 70 people would be more than one third of the yearly deaths, and that's in about three weeks only.

The mayor of Nembro: "Last year [in my city, between the 1st and 22nd of March] we had 14 deaths. This year, between 110 and 120."

The mayor of Scanzorosciate: this year we got 6x as last year [in the same period].

In Caravaggio 50 this year (only 2 officially of coronavirus) vs 6 last year.

In Stezzano, 40 people died in March so far (all with symptoms compatible with coronavirus) vs 10 in the same period last year.

In Selvino, 20 people died over the last two weeks. Before that, it took one and a half year to get the same amount of deaths.

In Coccaglio (Brescia), 5 people died of COVID but 24 more died in a residence, none tested. 36 people died between the 1st & 24th March, compared to a yearly average of 75 (link, via @Franceskamarel).

In Orzinuovi (Brescia) have been 75 deaths in the period 1st-26th of March 2020 versus 16 in the same period in 2019: more than four times as much. Of the 75, only 40 have been confirmed COVID-19.

In Manerbio hospital (Brescia) there have been 163 deaths in the period 1st-26th of March 2020 versus 42 in the same period in 2019.

Notably, in Selva di Val Gardena / Wolkenstein, a town which is not in Lombardy but in Bolzano province, 13 people died so far in March 2020 compared to 2 in the same period in 2019 (source: the necrologies). This is worrying as it would suggest that it’s not much that Lombardy’s mortality is high because of specific problems there, but (also) because other regions are underestimating direct or indirect deaths too.

I summarized the data in a table (which I will attempt to keep updated): link.

Could it have been just a bad year of flu?

A legitimate question.

I examined the data from Dalmine, one of the cities in the table (source). (Why only Dalmine? Because it’s the city whose historical data I found the easiest.)

As Dalmine’s population has been changing over the years, I normalized the number of deaths in any given year to the town’s population in 2018.

In the worst year of Dalmine’s recent history, “only” 19% more people than average died.

Could it be, though, that those excess people all died concentrated in the same month?

I took the total number of deaths in the worst year (219) and subtracted the average number of yearly deaths (184) to obtain the excess deaths (35.3).

I then divided it by the average number of deaths in an average month of an average year (15.3) to obtain a multiplier of 2.3x, which is still less than half the multiplier from COVID in March 2020 (5x).


Of course, the sample of towns is very small and their population is small too.

I didn’t cherrypick the data: I took all data I could find from towns in Bergamo or Brescia province regarding March 2020 data compared to March 2019. Of course, there is some selection bias: the journalists that dug up that data looked for it in the worst-hit towns.

Therefore, the conclusion to take out of this data is not that in every city Italian 5 times as much people died in March 2020 compared to March 2019. Instead, the two conclusions we can take from the data above is: (1) when coronavirus hits a town badly, it can cause lots of people to die and (2) at least in the worst-hit regions of Italy, deaths are under-counted.

What about testing?

Testing is completely inadequate. I recently wrote a piece with evidence (link).


This data is by no means complete; however, it paints a grim picture.

If any reader who lives in a city badly hit by the pandemic wants to do the same exercise, comparing deaths in his city in the periods March 2019 and March 2020, it would help a lot! Please share the results.

Many thanks to @Franceskamarel who provided me with 2 of the 4 sources used for this article.

Subscribe to this newsletter (it’s free, no spam whatsoever):

Share this post with your friends and colleagues:

Share Covid Stuff

Bergamo hospital (rendering before construction)