COVID symptoms: REACT study identifies new giveaways of infection

New data from around the UK could help us catch more cases
16 February 2021

Interview with 

Joshua Elliott, Imperial College School of Public Health


A stylised coronavirus particle next to a woman wearing a facemask.


Last year, a “holy trinity” of symptoms - cough, fever and sudden loss of smell and taste - were introduced as the cardinal giveaways that someone might have coronavirus infection and they should go and get a test. Since then, other symptoms have been documented, but they’ve not been regarded as adequately specific to warrant inclusion: too many people, it was felt, might suffer false alarms. But, at the same time, we’re currently missing a significant number of cases, and with the emergence of new variants of the virus, might the symptom list be in need of an update? Josh Elliott thinks so. He works at Imperial College’s School of Public Health...

Josh - One of the main reasons for doing this study was to find what were the most informative symptoms for whether or not you have COVID. Currently you are eligible for community testing if you have one of three symptoms: loss or change to your sense of smell or taste, a fever, or a new continuous cough. And our study, we considered loss of sense of smell and loss of sense of taste as different symptoms. We find that all of these symptoms are the most important symptoms for predicting your risk of COVID. However, there are additional symptoms that we identify which give extra information on whether you have COVID or not - your risk of COVID.

Chris - We can come to those in just a second, but can you tell us, first of all, how you actually did this?

Josh - We use data from a study called REACT 1 which stands for Real-time Assessment of Community Transmission, One. Roughly every month 160,000 people are included, which is a random sample of the English population five years plus. In every single individual, we have a PCR test result, we have information that they give us from questionnaires including about their age, their sex, their ethnicity, et cetera, as well as any symptoms. From that data we can make models which show which symptoms provide important information as to your risk of COVID.

Chris - What's interesting is that you've studied this across the time when we've seen the emergence of new variants of coronavirus in the country. So did you see any difference in what we'll call the classical coronavirus, the parent, and then the new virus we're now seeing?

Josh - So we indirectly studied the new variant by looking at a round that took place in January 2021, when over 80% of positive tests were due to the new variant versus a round which occurred at a similar seasonal time in November and December when the proportion of positives were only around 15 or 16% due to that new variant. And in doing this, we could find that loss of sense of smell was less predictive for the later round in January than in the earlier rounds. And also we found that the new persistent cough was more predictive in the January round than in the November/December round.

Chris - So basically with the new variant you cough more but you can taste your dinner better than you would have done had you had the existing parent virus. Why do you think that is?

Josh - It's difficult to speculate exactly. It might be something to do with the way that the virus interacts with the lining, where chemicals have attached to the lining of your nose so that you can smell, or it might be to do with the way that the virus levels are just higher in infection. It's not clear at the moment.

Chris - And what about the other symptoms that you were looking up? Cause obviously you're giving questionnaires to people, you're extracting a lot of data. So in addition to that magic number of four symptoms that you looked at, were there any other really good predictive symptoms that would steer us towards a diagnosis of coronavirus in people who had them?

Josh - Yes, and one key thing I want to stress about these symptoms that I'm about to list are that they provide additional information alongside those other four symptoms. So it's only when you consider all of these together that you're getting the most out of them. So those symptoms were across all ages: chills, headache in children - so people aged 5 to 17 - appetite loss in adults of any age, and muscle aches in adults aged between 18 and 54. So all of those are adding extra information to tell us about your risk of COVID based on what symptoms you have

Chris - With the list of symptoms that you've found are any emerging as ones that perhaps we should consider adding to the three or four cardinal features that we're using to trigger people to go and get a test? Because the government have been quite resistant about this. They were asked about this last year, should we extend the definition of what constitutes someone who might have coronavirus? And they felt that they shouldn't do that because it would end up leading to too many people suspecting themselves of coronavirus when in fact they didn't have it. From your data. Is there evidence that perhaps we should have more symptoms on that list?

Josh - Yeah. So of all the people that has any one of the four symptoms overall, you're only picking up about half of the overall number of symptomatic infections in the community. You would get from halfway to three quarters of all symptomatic cases in the community if you used our suggested models. In order to do that, you would have to offer tests to 2 and a half times the numbers of people in the community.

Chris - So in other words, the price to be paid for having a wider definition is we have to do more tests because we're going to drag in more people who haven't really got coronavirus. We have to test them to prove they haven't. But at the same time, the payback is we could, potentially, instead of missing half of cases, we'd miss far fewer positive cases because we would end up testing more people and finding them.

Josh - That's true.

Chris - What have the government said when you present them with this, are they likely to embrace this and to include your symptoms that you're identifying as a way to find more people? Because this has proved to be a real headache, hasn't it, tracking down, pinning down the virus because at the moment we miss too many cases and it makes it very hard to control.

Josh - Yes, I do think that we need to do something in order to reach more people that might have the virus and also coming forward and wanting testing, but currently aren't eligible. And I do think that there is interest from the government in updating the eligibility rules for testing in the community.


Add a comment