This Oxford modelling study has been getting a lot of attention, with suggestions that it implies there has already been substantial amount of COVID-19 infection in UK. But it's important to take a look behind the media headlines... dropbox.com/s/oxmu2rwsnhi9… 1/
The analysis focuses on deaths in UK and Italy. The challenge in any outbreak is that multiple scenarios might be consistent with a timeseries of deaths viewed in isolation. Are there lots of infections and a low severity? Or fewer infections and a higher severity? 2/ 
The authors suggest that multiple scenarios could be consistent with the number of deaths reported, including ones in which there have been a lot of infections. Their figure below shows some possibilities. But remember, this is just focusing on the death data... 3/
There are several (admittedly imperfect) other datasets out there that have suggested a reasonable proportion of infections might be symptomatic, particularly in older groups (and hence severity is high)... 4/
Data from testing on the Diamond Princess (which was unfortunately not systematic) suggested around half of cases were asymptomatic niid.go.jp/niid/en/2019-n… 5/
On evacuation flights, there were a mix of symptomatic and asymptomatic passengers. Very small numbers here, but no clear evidence that many were asymptomatic. imperial.ac.uk/media/imperial… 6/
This contact tracing study also identified some asymptomatic infected contacts, but not a large proportion. medrxiv.org/content/10.110… 7/
Of course, all these studies have limitations for estimating asymptomatic infections, and to understand how much infection there has actually been, what we really need is population serological surveys. This is the key message of the Oxford paper (and one I fully agree with) 8/
The Oxford group have previously done some great work on immunity dynamics (e.g. nature.com/articles/s4146…), so am looking forward to seeing the results of their COVID-19 serosurveys. 9/9
