The coronavirus pandemic has unleashed a deluge of statistics. In the space of a couple of months the public has been hammered with once-obscure figures that are now of immense importance to our lives. Discussions about case fatality rates, reproduction numbers and ICU surge capacity have become part of the fabric of life under lockdown.
On March 24, one particular figure lifted itself out of this background hubbub. A story in the Financial Times claimed a University of Oxford study concluded that half of the UK population may have already been infected with coronavirus. The story quickly did the rounds of UK newspapers, getting picked up by the The Times, Daily Express, Evening Standard, Daily Mail and The Sun, each story echoing the hopeful news that half of us had already had coronavirus and that herd immunity is around the corner.
These headlines are way off the mark. The Oxford study – which has not been published in a scientific journal or scrutinised by other scientists yet – offers a set of hypothetical situations about the possible extent of coronavirus transmission in the UK. The headline statistic, that more than half of the UK population has already been infected with coronavirus, is not supported by real-world data, epidemiologists argue.
“It’s a little concerning that they’ve taken it straight to the media,” says Tim Colbourn, an epidemiologist at University College London’s Institute for Global Health. “It has not been properly sense-checked against any data.” The authors of the Oxford study did not respond to WIRED’s request for comment in time for publication.
The study, led by Sunetra Gupta and José Lourenço at the University of Oxford’s Department of Zoology, puts forward several hypothetical scenarios about the spread of coronavirus in the UK. In the most extreme scenario they estimate that if the virus had started being transmitted 38 days before the first confirmed death then 68 per cent of the UK population would have been infected by March 19.
But this modelling rests on an improbable assumption: that just one in every 1,000 people infected with coronavirus will need to be hospitalised. This assumption just doesn’t match real-world data, says Colbourn. “We can already see just by looking at Italy […] that that figure has already been exceeded,” he says. In Lombardy – despite the region being under lockdown since March 9 – more than one in every 1,000 of the entire population have already been hospitalised due to coronavirus. According to the most recent data, the death rate is currently around 0.42 per 1,000 people.
“The fact that they didn’t look at that data is extremely concerning given the headlines it’s now generating,” Colbourn says. His concerns are echoed by seven public health academics who released statements to the Science Media Centre. “This theoretical simulation rests on a key assumption which may be or may not be correct,” said James Naismith, director of the Rosalind Franklin Institute, in his statement to the SMC.
Although the headlines generated by the Oxford paper may be well wide of the mark, at its heart, the paper makes a simple point: we need to know how many people in the UK have already been exposed to coronavirus. Working this out will require serological tests – blood tests which look for the presence of antibodies that indicate if someone has already been exposed to a disease. In a press conference on March 24, health secretary Matt Hancock confirmed that the UK has ordered 3.5 million of these antibody tests to work out how many people have already contracted coronavirus. The tests the UK has purchased will soon be going through a validation process before being sold to the public.
Until then, we just won’t know the true proportion of people who have contracted the disease without showing any symptoms, but it is likely a much lower number than the Oxford study assumes. An analysis of Chinese cases as well as data from repatriation flights found that 40 to 50 per cent of infections were not identified as cases. Data from the Diamond Prince cruise ship found a similar level of asymptomatic cases, but without very widespread testing in more general populations we simply don’t know how many people contract coronavirus without showing symptoms.
But public health experts are concerned that the headlines generated by the Oxford study will lull us into a false sense of security. “It said something that fits with what everyone wants to hear, which is ‘we don’t need to have lockdown because everyone already has it’”, says Devi Sridhar, professor of global public health at the University of Edinburgh. “As much as I really want to believe that all the evidence we’re seeing from other places says it’s not [the case].”
Others are worried that it might undermine the government’s social distancing measures, put into place after modelling from Imperial College London suggested that the NHS was likely to be overwhelmed by the rate of hospitalisations due to coronavirus. If people believe that they have already been infected with coronavirus, they may be less likely to follow the government’s social distancing regulations.
“I am surprised that there has been such unqualified acceptance of the Imperial model,” the study’s supervisor, Sunetra Gupta, told the Financial Times. But in an evidence session before the Science and Technology Committee the lead author of the Imperial study, Niall Ferguson, said that the current data did not support the scenarios in the Oxford paper. “We don’t think [the model] is consistent with the observed data,” he says. The Imperial model took into account studies in some Italian villages where every single resident was tested, giving a much more robust idea about how common infections with coronavirus were. “Those data all point to the fact that we are nowhere near the [Oxford study] scenario in terms of the extent of the infection.”
One part of the paper is well worth paying attention to: we need to know how many people have been exposed to coronavirus. To that end, thousands of at-home finger prick tests could be sent to people self-isolating with symptoms, The Guardian reports. Once those results start coming in, we might have our first indication of how widespread coronavirus truly is – but not before then.
Matt Reynolds is WIRED’s science editor. He tweets from @mattsreynolds1
Coronavirus coverage from WIRED
😓 How did coronavirus start and what happens next?
❓ Does alcohol kill coronavirus? The biggest myths, busted
🎮 World of Warcraft perfectly predicted our coronavirus panic
✈️ Flight data shows the huge scale of Covid-19