The number of people who died from Covid-19 is almost double the number we hear announced every day.
On May 1, the number of coronavirus deaths announced by the UK government was just over 28,000.
Looking at the death records presented at the time, the number is higher: just under 36,000 death certificates mentioned in Covid-19.
The measure preferred by statisticians, counting all deaths above what would be expected, was even greater: more than 50,000.
Each measure answers different questions.
Why shouldn't I trust the daily government figure?
Every day, the Department of Health and Social Assistance (DHSC) reports the number of coronavirus deaths that have been reported across the UK.
This is the figure read at the daily press conference and the one used on most international comparison sites.
But that includes only deaths of people who test positive for coronavirus.
This is good for scientists who want to monitor the epidemic's growth patterns accurately: the rate of increase and the time spent at the peak are useful measures, says Prof Sylvia Richardson, president-elect of the Royal Statistical Society. Cambridge University.
But it is a bad measure of the total number of deaths because it misses people who have never been tested.
When testing was largely limited to hospitals in the UK, these daily numbers were missing from most deaths in the community.
In addition, different countries use different definitions.
For example, England excluded deaths outside hospitals from its daily count until a few weeks ago. Scotland, Wales and Northern Ireland do not.
And Belgium includes suspected cases of coronavirus in its daily count, which makes its numbers appear extraordinarily high compared to other countries.
This makes accurate comparisons between countries difficult, and scientists warn against reading too little absolute differences in these daily statistics.
When each country counts things differently, statisticians resort to a different measure – with a simpler definition.
Looking at all the deaths
If you look at all the deaths in a country, regardless of the cause, you will capture the deaths missed by laboratory tests, the deaths misdiagnosed and those caused by the strain that the virus puts on our society.
Obviously, you will catch heart attacks and car accidents that may have happened anyway.
But the total number of deaths recorded in a week usually follows a predictable pattern.
It has skyrocketed since late March, well above the previous weeks or what would be expected at that time of year. That number has dropped in recent weeks, but we are still seeing more deaths than would be expected at this time of year.
And it is these extra or "excessive" deaths, the difference between the number we normally see and what we are seeing at the moment, that statisticians use to capture the true price of coronavirus.
What does that tell us?
Increasing the weekly excess during the weeks of the epidemic, it reaches just over 50,000 by May 1, higher than the number of death records that mention Covid-19 or the UK government's daily count.
About three-quarters of this excess can be explained by deaths that mention Covid-19 on the death certificate – that total was just over 36,000.
But there are still nearly 14,000 deaths, a significant increase, which could be undiagnosed deaths, caused directly by the coronavirus or indirectly.
This provides a clearer picture of the total cost of the virus.
It can also be a better measure to compare countries because it does not depend on which deaths are counted.
But it still "needs to be put into perspective," says Professor Richardson.
The UK releases death data every week, but this is not the case for all countries.
Some publish data daily, others monthly or quarterly, making comparison difficult from a similar starting point.
The things that make coronavirus dangerous also differ between countries.
You would expect to see more excess deaths in Italy, where the average age is 47, than in Ireland, where it is 40.
That's all before you take the actions that governments take to deal with the virus: how quickly and forcefully they block or how effectively they test and track and quarantine cases of the virus and how the health system handles it.
Without that extra qualitative information about how these measures are taken "you could miss something very obvious", warns Professor Richardson.
"You need to understand the social context of the measures and how they were implemented."
The calculation of who was most affected alone does not indicate why they were the most affected.
And it just tells you how countries are doing now.
We will not have a complete picture of who has had the worst epidemic until countries go through it.
There will still be deaths from second peaks or damage from any blockages.
The lack of a single number at the moment that gives the definitive image of who was the most affected or who is dealing better does not mean that we should avoid all comparisons.
But, says Professor Richardson: "You need several numbers – and some understanding."