Coronavirus statistics: what can we trust and what should we ignore?

Input square image here

12th April 2020


on The Guardian

Sylvia Richardson, director, MRC Biostatistics Unit;
David Spiegelhalter, chairman, Winton Centre for Risk and Evidence Communication

The flurry of figures, graphs and projections surrounding the pandemic is confusing. Two experts guide us through the maze
The past few weeks has seen an unstoppable epidemic … of statistics. The flood threatens to overwhelm us all, but what do all these numbers mean? Here are eight statistics you may see, with some warnings about how much we might trust them.

1 The number of new cases each day This can be a very poor reflection of the number of people who have actually been infected, as it depends crucially on the testing regime – up to 9 April, 1.3 million tests had been carried out in Germany, versus 317,000 in the UK.

2 The number of new deaths each day The range of sources is bewildering. The daily announcements should be treated with caution as they only include deaths in hospital of those who have tested positive for coronavirus, and there is generally a delay in reporting deaths of a few days or even longer. For example, while on 27 March the government announced that 926 Covid-19 deaths had so far taken place in English hospitals, NHS England now reports that the true figure was 1,649. The gold standard is the number of death certificates collated by the Office for National Statistics: it report at least 1,568 mentions of Covid-19 for all deaths up to 27 March, but this will increase as registrations come in.

3 The total number of deaths Graphs of accumulating deaths are shown at the daily government press conference, but they are a hopeless tool for spotting trends: we need daily counts to see whether we have reached a plateau (which will not be a “peak”). 

Read the full article on The Guardian