select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
A preprint, an unpublished non-peer reviewed study, reports on the latest data from the REACT-2 study looking at persistent symptoms following SARS-CoV-2 infection.
This Roundup accompanied an SMC Briefing.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This is an important, large-scale, study that emphasises the importance to the nation’s health of long-lasting symptoms after Covid-19. According to this preprint from the REACT-2 study from Imperial College, the scale of the problem is quite alarming. The researchers estimate that about 1 in 17 adults (aged 18+) in the community population of England would report that they have had one or more Covid-19 symptoms lasting at least 12 weeks after they started. Not 1 in 17 of the people who have had Covid-19; they mean 1 in 17 of the whole adult population. That’s well over 2 million people. The researchers estimate that over 1 in every 3 people who had a Covid-19 infection, that was symptomatic initially, went on to have at least one of the symptoms lasting 12 weeks or more. The results can’t tell us clearly how serious those symptoms were, in terms of their effects on the patients’ lives, because the researchers asked only about severity when the person had Covid-19, and this is taken to refer to the initial disease, not what might be going on 12 or more weeks later. Also some of the symptoms may not be very serious. But some of them certainly are, and these results clearly point out how important and, in many cases, debilitating these long-lasting symptoms are, and how vital it is to understand them properly and to provide adequate treatment and support services for the people involved.
“I mentioned the need for better understanding of long-lasting symptoms after Covid-19, because another thing that this study draws attention to, alongside other studies, is how difficult it is to define, and to measure the prevalence of, the condition or conditions involved. The REACT-2 results come from a survey of a reasonably representative sample of adults in England, who were surveyed primarily in order to estimate how many of them were positive for antibodies to the virus that can cause Covid-19. ONS have also been publishing results1 about ongoing symptoms after a Covid-19 infection, also based on a survey of a reasonably representative sample of adults in the whole UK (including England, of course), who were surveyed in the Covid-19 Infection Survey (CIS) primarily in order to estimate how many of them had a current infection with the virus. ONS estimated that about 741,000 people in England had at least one self-reported symptom lasting at least 12 weeks; this REACT-2 preprint estimates that well over 2 million people in England had one self-reported symptom lasting at least 12 weeks. So REACT-2 estimates that there are well over three times as many people as ONS did, with at least one symptom persisting that long. The main reason for that discrepancy is that the two studies are estimating slightly different things. ONS were estimating the number of people who had long Covid (defined in terms of symptoms lasting at least 12 weeks) on a particular date (2 May), while REACT-2 is giving a measure of how many people ever had long Covid (defined in terms of symptoms lasting at least 12 weeks) at some time over three rounds of their survey between September 2020 and February 2021. You’d expect that to be higher than the ONS count of people who actually still had the condition on a specific day – some of the people who had long Covid at the start of the REACT-1 period would have recovered from it by the end. But the REACT-2 estimate is still higher, in terms of the chance of Covid symptoms lasting longer in people who have had Covid, as the REACT-2 preprint makes clear.
“The REACT-2 preprint mentions several other studies, carried out in the UK and elsewhere, that show a bewilderingly wide range of estimates of how common this condition is, though most of them used different methods from the surveys used in REACT-2 and the CIS. I’ll concentrate on REACT-2 and the CIS because they both used population surveys, in the same country, and based their findings on the self-reported answers that the participants gave to questionnaires. So why is there such a big difference? One possibility is that the surveys were carried out a different times and weren’t estimating exactly the same thing anyway. But another important set of reasons relates to exactly what was asked, and how. The REACT-2 preprint includes quite a lot of discussion of this. I’ll mention just some possible reasons for the difference.
“One reason is the symptoms that the respondents were specifically asked about. In REACT-2 there was a list of 29 of them, ranging from things like a runny nose to severe fatigue (e.g. inability to get out of bed), plus the possibility of writing in a symptom that wasn’t already listed. In the CIS the list includes fewer symptoms, 21 of them. Interestingly, the CIS list omits some of the symptoms on the REACT-2 list that, superficially at least, appear a bit less severe, such as runny nose and sneezing, but the CIS list contains symptoms like worry, confusion, low mood, and difficulty concentrating, that are not on the REACT-2 list in any form. (REACT-2 respondents could have written them in, and the preprint does report that “brain fog” was quite often written in.) Another issue of the detail of what was asked is that, in the REACT-2 questionnaire, no mention is made of long Covid – instead respondents were asked whether they had any of the listed symptoms, and, if so, how long they lasted. In CIS, the questions about the list of 21 symptoms are asked explicitly in relation to long Covid. So someone who still had one or more of the symptoms that are on both the REACT-2 list and the CIS list 12 weeks after their initial Covid-19 illness, but didn’t consider that they had long Covid, would probably not be counted as having long Covid, but the same person responding to REACT-2 would probably count towards their estimate of people with a symptom lasting 12 weeks or more.
“That’s a reason why the CIS figure could be lower than the REACT-2 figure, but there are potential reasons for difference that go the other way too. Because REACT-2 basically asked how long each symptom went on from the time the person first had Covid-19, it could miss some people whose symptoms came and went (as has been quite frequently reported by people with long Covid). Because CIS specifically asks about long Covid, it might be more likely to pick up such people. Various other possible limitations of the REACT-2 findings are mentioned in the preprint. And a potential issue for both surveys is that no comparison is made with people who said they did not have Covid-19. Some of the listed symptoms on both lists arise pretty commonly in people who have never had Covid-19, because they can be caused by many different illnesses. So it’s possible that the figures in both surveys are over-estimates of the persistence of Covid-19 symptoms, and arguably the question wording on the CIS questionnaire could be less liable to that issue.
“These differences are important, I’d say, not so much because they produce very different estimates of how common long Covid might be, but more importantly because we need a good understanding of what’s actually going on, in terms of symptoms and the people who have them, if we’re to develop a good system of support and treatment for the people who are affected. And we can’t have that understanding without a bit more clarity on what the condition is, or, more accurately, what the conditions are. (The REACT-2 preprint analyses the patterns of symptoms that people reported, and finds more than one pattern, and previous researchers and patients have described that there may well be several different conditions involving persistent Covid-like symptoms.)
“Finally, the lack of clarity on exactly how many people are affected shouldn’t draw attention away from the fact that the lowest estimates still show a huge amount of largely unmet need. This REACT-2 preprint adds to the evidence for that, and indeed adds some detail that could well be important in planning to meet that need.”
Dr Julian Tang, Honorary Associate Professor/Clinical Virologist, Respiratory Sciences, University of Leicester, said:
“Both the REACT study below and the Nature Medicine paper from Norway (https://www.nature.com/articles/s41591-021-01433-3) show that a significant proportion (30-50%) of adults with COVID-19 develop persistent ‘long COVID’ symptoms that last for at least 3 months – which can include: fatigue, muscle aches, chest pain, poor concentration, shortness of breath, loss of taste and smell.
“This knowledge and experience has been accumulating over the past 6 months – as the vaccine rollout has reduced the number of hospitalisations and deaths, the spotlight is switching to long COVID, which may become more of a burden on community healthcare services, like GPs.
“In learning to live with this virus, with COVID-19 vaccines that are remarkably effective at preventing severe disease, hospitalisations and deaths, the longer-term healthcare burden may lie in dealing with long COVID symptoms that may be similar to other post-viral syndromes that we see with glandular fever, for example.
“We don’t see this with seasonal flu, but flu has been with us for much, much longer. In the earlier 1918 Spanish flu pandemic, we saw unusual complications like ‘encephalitis lethargica’ (the subject of the 1990 film ‘Awakenings’ with Robert De Niro and Robin Williams), which we do not see nowadays. So these long COVID symptoms may be a feature of this new virus infection in this new human host – which may become less common as we develop a longer and more diverse immunity to this virus.
“In the meantime, prevention is better than cure, so rolling out COVID-19 vaccinations to all global populations to reduce COVID-19 infections and the incidence of any related complications, will help all of us to reduce the cost and ‘carer’ burden for these long COVID cases – especially in Western Caucasian populations that seem to be more effected by long COVID.”
Comment on both this preprint and the CONVALESCENCE study on risk factors for long COVID, which is under the same embargo:
Dr David Strain, Senior Clinical Lecturer, University of Exeter, said:
“These two separate studies, both tell a similar story – Long COVID is substantially more common than we originally thought, the reported number from the REACT-2 study being around double the number estimated by the ONS report earlier this month. It is also having a substantial impact on people of all ages with lasting consequences. Although the risk in the younger population is smaller, it remains significant, which is of concern giving the rising cases of the Delta variant of COVID in the, as yet, unvaccinated under 30s.
“Both of these report on large population samples recruited and identified through different strategies, yet show remarkably similar associations with gender, ethnicity and pre-existing diseases. This gives vital information beyond simple risk assessments, allowing researchers to start exploring potential mechanisms of the disease in the search for potential treatments or cures. They also inform clinical service, as the number of long Covid assessment clinics rise beyond 80 across the UK. If the estimated 2 million, and climbing, sufferers are to be seen within the NHS this will require significant investment into both the education of health care practitioners, who are now treating a condition that did not exist 18 months ago, but also to provide the service in the community and secondary care bases.
“As with all survey or record based studies these both come with the caveat that it is dependent on people with lived experience of the condition reporting their symptoms to their health care practitioner. It has previously been suggested that this reporting bias may have accounted for some of the ethnic disparities and gender inbalance, however the similar results using two different strategies is highly suggestive that these are true findings.”
Preprint (not a paper): ‘Persistent symptoms following SARS-CoV-2 infection in a random community sample of 508,707 people’ by Matthew Whitaker et al. was posted online at 00:01 UK time on Thursday 24 June 2021. This work is not peer-reviewed.
All our previous output on this subject can be seen at this weblink:
www.sciencemediacentre.org/tag/covid-19
Declared interests
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. I am also a member of the Public Data Advisory Group, which provides expert advice to the Cabinet Office on aspects of public understanding of data during the pandemic. My quote above is in my capacity as an independent professional statistician.”
None others received.