Can we trust the Oxford study on Covid-19 infections?

Earlier this month, the MSc students I teach at the London School of Hygiene & Tropical Medicine ran an outbreak investigation as part of their epidemiology assessment. They were told several people had fallen ill, and it was their job to piece together various fragments of information – about symptoms, timings and locations – to work out what might have happened.

We face the same challenge with every outbreak: we never have a single perfect dataset early on. Instead, we have to combine the information we do have to work out what might really be happening.

This week, a preliminary study from a team at the University of Oxford gained a lot of media attention, with headlines suggesting it implied the majority of people in the UK might have already been infected with Covid-19. Few would dispute that there are more cases in the UK than already reported; our group recently published a rough analysis that suggested only 5–7% of people who’ve fallen ill with Covid-19 have been recorded as confirmed cases in the UK data. But how plausible is it that there have been millions of infections out there that were very mild (or symptomless) and went undetected?

First, let’s look at what the Oxford team actually did. Its analysis focused on deaths in the UK and Italy. If we view the data in isolation, there are several scenarios that could potentially be consistent with the observed pattern of deaths. Perhaps there are lots of infections and a low severity. Or fewer infections and a higher severity.

Coronavirus: the week explained – sign up for our email newsletter
Read more
The Oxford researchers suggested there were multiple scenarios that could explain the deaths reported in the UK, including those in which there have been a huge number of infections that haven’t been detected. The media picked up on the most extreme of these scenarios, which implied the majority of the UK had already been infected. But remember, this analysis focused solely on deaths – and that’s not the only data we have for Covid-19.

There are now several datasets that suggest many infections come with symptoms, particularly in older groups, and hence it is less likely that there are vast numbers of infections going undetected. Such datasets include testing on the Diamond Princess cruise ship, testing of passengers on evacuation flights, the age pattern of Covid-19 cases, and infections detected among contacts of cases.

Based on these studies, it seems about 20–80% of people infected with Covid-19 could show symptoms. If this range turns out to be correct – and we combine it with our estimate that one in 15 people with symptoms are being reported – it would mean that hundreds of thousands of people in the UK have probably been infected with Covid-19 already, but not tens of millions. Of course, all these early studies have limitations in how the data was collected and what groups were being studied. To understand how much infection there has actually been, what we need are studies that collect blood samples from the wider population and test for previous exposure to Covid-19. As it happens, this was the key point made by the Oxford researchers in their paper, and it’s one I fully agree with.

UK healthcare workers – the coronavirus protection they need and what they’ve got
Read more
There is growing evidence that, on average, people who show Covid-19 symptoms have a 1–1.5% risk of death. This has been estimated in studies of data from Wuhan, early international cases, and the Diamond Princess (with data adjusted to account for the older age of the cruise ship passengers). But this 1–1.5% risk just tells us what happens to people who have clear symptoms. If, as the above studies suggest, only 20–80% of infections come with symptoms, it would mean that for every 100,000 people who get infected with Covid-19, we would expect somewhere in the region of 200–1,200 deaths (ie between 100,000 x 20% x 1% and 100,000 x 80% x 1.5%).

We still don’t know the exact risk, and we won’t until we have the results from a large-scale analysis of blood samples. But the early pieces of information we do have suggest that this is an infection that can cause a large amount of severe disease, and the majority of the UK population are still susceptible to it. Until we have evidence to the contrary, this is the scenario we have to plan for.

Writing about COVID-19

As well as building up a resource of information and analysis on COVID-19, we want to ensure that we pass on any tips about what can go wrong when writing about this subject: and how to get it right! If you have experience in writing about this area and feel you have advice that would help others, please contact us at: covid19editor@wfsj-briefing.org.

We’d also like to hear from you at that address if you would like to regularly contribute links to the site. If you just want to suggest links on an occasional basis, please send them to: covid19@wfsj-briefing.org.