“19m Brits ‘have had virus’ & recovered”
The Daily Mirror front page, 15 May 2020
“Almost 1 in 3 Brits may have already had coronavirus – and most had NO symptoms, experts find”
The Sun, 14 May 2020
“Coronavirus: 19 million people in UK ‘likely to have been infected already’, study finds”
The Independent, 15 May 2020
A number of newspaper headlines on 15 May reported the findings of a study from the University of Manchester. Readers have written to ask us if it’s true.
The newspapers describe a new research paper published in the International Journal of Clinical Practice. It says that: “unreported community infection [with Covid-19] may be 150 times higher than reported cases, providing evidence that by the end of the 2nd week in April, 26.8% of the [UK] population may already have had the disease and so have increased immunity” (A previous version of the paper gave the 29% figures used by the newspapers.)
In their announcement of the research, the University of Manchester said: “Over 25% of the UK likely to have had COVID-19 already”. The statement begins by saying that, "A team of researchers from The University of Manchester, Salford Royal and Res Consortium, have shown that a significant proportion of people in the UK - over 25% - is likely to have been infected already by the COVID-19 virus.”
If the journal paper’s findings were correct, 29% of the UK population would amount to roughly 19 million people, so the newspaper headlines described the research accurately.
However, several other scientists have publicly criticised the study, saying that they believe its methods do not provide a reliable way of estimating how many people have already caught the new coronavirus.
Be first in line for the facts – get our free weekly email
What the researchers actually did
The researchers behind this paper did not directly test anybody, either for the new coronavirus, or for any antibodies that would show they had been infected in the past. Instead, they used a statistical model to estimate the true number of cases that would have existed in the real world, based on the changing number of cases that were reported.
This involved making some uncertain assumptions, including how long it takes for an infected person to become infectious to other people. The paper also mentions as a weakness its assumption that a consistent percentage of all the real cases would be tested and reported. There are reasons to expect that this might not be true, if cases were identified more successfully at different times, or in different places.
Mike Stedman, one of the researchers, acknowledged that “The figures are not perfect, with the numbers of severely ill patients as a proportion of the total cases being used as a market [it’s possible this is a misprint for “marker”] for estimates of wider infection.
“Only extensive antibody testing could give us a more accurate picture - but as that is only just becoming available, we believe this form of modelling is important in informing the best approach to unlocking the population”.
What are the criticisms of the research?
Carl Bergstrom, a professor of biology at the University of Washington, has written in detail on Twitter about why he believes that this research is not reliable, saying he found its statistical method “bizarre”.
The study has also been criticised for several reasons by Dr Konstantin Blyuss, Reader in Mathematics at the University of Sussex, and Adam Kucharski, Associate Professor in Infectious Disease Epidemiology, London School of Hygiene & Tropical Medicine.
Dr Blyuss said: “There are several problems with this study. The first concerns the data that have been used, namely, the data analysed in this study only covers the period up to 23rd April. On the basis of those limited data, the study suggests the peak of the number of daily new confirmed cases occurred on the 8th April 2020, and the numbers of new cases has been below 4,000 since that data. Both of those conclusions directly contradict what the current data show, with the numbers of new cases exceeding 4,000 for most of April and the first week of May.”
Dr Kucharski says, “It is likely that there is huge uncertainty in the estimates produced by the model used in this paper, and unfortunately this uncertainty is not reflected in the single value quoted in the paper and the press release.”
How many people have been infected?
We can’t know for sure yet. To illustrate the range of different numbers being put forward, another piece of research, also based on statistical modelling, from the MRC Biostatistics Unit at the University of Cambridge estimates that about 6.5 million people had been infected with the new coronavirus in England by 10 May. That would be about 12% of the population, which is much less than the 26.8% (or 29%) estimated by the Manchester study.
The most definitive results will eventually come by directly testing large numbers of people for antibodies, which would show who had been infected with the virus, whether they had symptoms or not.
When he was asked about the UK’s level of past infection by the Health and Social Care Committee on 5 May, the Chief Scientific Adviser, Sir Patrick Vallance, said, “The latest [antibody test results] I have seen is probably from just over two weeks ago, and reflects three weeks before that, because that is roughly how long it takes you to get your antibodies. It suggests something like 10% antibody positivity in London and something much closer to 3% or 4% in other places. It may be lower still in other places. I would expect to see variation in that across the country. We will get more information as those studies go on, but I would not expect to see antibody levels much above the mid-teens; but we will see.”
During the government’s daily press conference on 21 May, the health secretary Matt Hancock said: “We have now had the results of our antibody surveillance study. This has told us that around 17% of people in London, and around 5% or higher in the rest of the country, have tested positive for coronavirus antibodies.” Mr Hancock did not give an overall percentage for the UK, but these results suggest a figure considerably lower than 26.8%.