select search filters
roundups & rapid reactions
before the headlines
Fiona fox's blog

expert reaction to new ONS data on prevalence of ongoing symptoms following COVID-19 infection in the UK: 4 June 2021

The Office for National Statistics (ONS) have released estimates of the prevalence and characteristics of people with self-reported “long COVID” using data from the UK COVID-19 Infection Survey.


Dr David Strain, Senior Clinical Lecturer, University of Exeter, said:

“These figures show the devastating longer term impact of COVID. The numbers of people with symptoms beyond 12 weeks have risen by almost a quarter, and nearly 200 thousand unable to work and function normally. With over 300 thousand more people reporting symptoms lasting beyond 12 months, it is becoming clear that this is not getting better on its own stressing the need for further research into treatments for this condition.

“This data emphasises the importance of vaccination at the moment, with the highest proportion of people suffering long COVID being those in the demographic that are currently being called for their vaccination. Our recent work has demonstrated that the majority of people suffering with long COVID benefit from vaccination, and clearly the best way to prevent long COVID is to not get COVID in the first instance.”  

“These data are self-reported therefore come with all of the caveats of how people experiencing a condition may interpret their symptoms. One person’s severe pain may be another’s mild ache. However it is also true that an individual is the best person to assess the impact of their own symptoms, and whether someone feels they can return to work, or manage a household will be reflected in these scores.  The self-reported nature may also account for the apparent higher prevalence in the younger, fitter populations, as these people are more likely to report and externalise a significant dip in their ability to perform day-to-day activities, whereas the more mature may be more likely to accept a disproportionate drop in their function attributing it to the lockdown or ‘their age’. In this case the true prevalence could be much higher.”


Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“Like the previous ONS release on long Covid* from the beginning of April this year, this new bulletin and data release is an important contribution to our understanding of what’s going on. One reason that it’s important is because the data come from a survey of a representative sample of people across the UK community population, so the results should represent what’s actually going on across that population. Because the condition is a new one and because treatments and medical interventions for it aren’t universally available or agreed, data based on health service sources are likely to be very incomplete, and using a survey that represents the whole population can get round that limitation.

“The headline results are quite dramatic. ONS estimate that a million people living in private households in the UK would report that they have long Covid symptoms persisting at least four weeks after an initial infection, that weren’t explained by anything else. That’s slightly down on the 1.1 million estimate from the last ONS bulletin, which covered surveys up to 6 March. But it’s still a very high number – roughly 1 in every 60 people. The estimated number who have had long Covid for at least 12 weeks is 869,000, or about 1 in 75 of the population. That’s quite a lot higher than the March estimate of 697,000. I think that’s probably just because more time has elapsed since the peak number of infections at the start of 2021, so that 12 weeks since their first infection will have elapsed for more people who were infected during the second wave. Being ill for 12 weeks or more is already a pretty unpleasant prospect, but the new estimate for the number who have continued symptoms a whole year or more after their first infection is 376,000, or about 1 in 170 people in the population. That’s a huge increase on the March estimate, though that’s not really surprising. Not many people had been infected before 6 March last year, so wouldn’t have been at risk of long Covid at the time of the previous bulletin, but the new data go up to 2 May, and people who were infected during that terrible first wave last Spring would now be at risk of long Covid lasting 12 months or more. 376,000 is very roughly the population of the entire urban area of Coventry, or Bradford, or Cardiff, so we have a whole Coventry-full of people who have had symptoms related to Covid-19 for a year or more.

“These long-lasting symptoms are certainly not trivial. Overall, of all the people with symptoms that lasted at least four weeks, 64% said that the symptoms were limiting their activity, and about a fifth (19%, or 192,000 people) said they were limiting their activity a lot. Those numbers don’t vary all that much with the length of time people have had the symptoms. And many people who didn’t say that their activities are limited will still be suffering pain and distress. It’s worth remembering that the definition of being a disabled person under the Equality Act is that you have a physical or mental impairment that is “substantial” and “long-term” in its effect on your activities. About 1 in 200 people aged between 35 and 70 has a new condition related to Covid that has a substantial effect on their activities (in that they said their activities are limited a lot), so potentially a new disability. In some cases, this long Covid potential disability won’t last long enough to become “long-term”, because that is taken to mean 12 months or more, and also in some cases, the person affected would already be disabled because of some other condition – but it’s still, potentially, a lot of extra disability.

“ONS report some quite marked differences between different population groups in the prevalence of these long-lasting symptoms.  The rates are highest in people aged 35-69, females, those living in the most deprived areas, those working in health and social care, and those with pre-existing health conditions.  However, it’s not easy to interpret those figures directly, because those factors are associated with one another in various ways. For example, age is associated with the chance of having pre-existing conditions, so if a particular age group is more likely to have long Covid symptoms, that might be directly because of their age, or because of the pattern of pre-existing health conditions, or something else entirely. To try to sort out the separate possible effects of factors like this, ONS carried out an analysis (logistic regression) in which the association between a factor, such as age, and the chance of having long Covid is adjusted to allow for differences in other factors. This adjustment can’t be perfect, so we still can’t know from the adjusted analyses what is causing what – that’s impossible in working with observational data such as these. But it might give a bit more of a clue. In addition to one stage of adjustment, the ONS statisticians also carried out an adjusted analysis only using data from people who had had a Covid-19 infection confirmed by a test. That analysis isn’t affected by the chances different groups have of being infected in the first place.

“Those adjusted analyses do indicate that the chance of having long Covid is, broadly, higher in people of working age than in people of school age and in people aged 70 or over, and in women than in men. Risks are higher, even after the adjustment, for people living in more deprived areas, though differences in risk between ethnic groups are small enough to be explained by statistical variability alone. The risk of having long Covid in children of primary school age or younger is quite low – but there are still 11,000 children in that age group who had symptoms lasting longer than 4 weeks, and 19,000 children aged 12 to 16. Children are still at risk of long Covid, even though, it would seem, at smaller risk than adults, particularly if they are 11 or under. The higher risks for people working in health care still show up after the first adjustment, but are no longer there if the analysis is restricted to people with a confirmed Covid-19 diagnoses from a positive test. This might well indicate that the higher risk in health works across the whole sample is because they had a higher risk of being infected, rather than anything to do with other characteristics or with what happened after infections – but nobody can be sure that that interpretation is right, because the survey can’t directly tell us about cause and effect.

“One thing that’s new in this bulletin, compared to the previous one, is that data are given on the specific symptoms that people reported, from a list of 21 symptoms. The three most common are fatigue (weakness or tiredness), shortness of breath, and muscle ache, but not far behind, the fourth is “difficulty concentrating”, which might be part of what people have described as “brain fog”. About 1 in every 140 people in the community population said they had fatigue, that they attributed to Covid-19, that had gone on (though not necessarily continuously) for at least 12 weeks. The other symptoms weren’t so prevalent, though for instance about 1 in 260 of the population said they had difficulty concentrating, over at least 12 weeks, that wasn’t explained by anything over than Covid-19. It would have been interesting to compare that with how many people said they had difficulty concentrating when they first had Covid-19, but we can’t do that because people aren’t asked about difficulty concentrating (and another 8 of the 21 symptoms) when they are first infected.

“In fact, though the data on symptoms is useful, some care is needed in interpreting it. Another issue is that we don’t know how many people, who hadn’t had Covid-19, would have reported these symptoms anyway. All the listed symptoms have many possible causes, and might have arisen for reasons other than Covid-19. In the previous (March) ONS bulletin, there was a comparison of reported symptoms between people who had definitely been infected and another, control, group that had almost certainly not been infected. That comparison is not included this time, on the stated grounds that the symptom list for making the comparison doesn’t include all the relevant symptoms. “That it doesn’t include all the symptoms seems to be an unfortunate consequence of the fact that not all the relevant symptoms had been recognised when the ONS Covid-19 Infection Survey (CIS) was set up a year ago. People were, and still are, asked only about a list of 12 specific symptoms in the section of the questionnaire that asks about initial symptoms when someone is first recorded as infected – though they can also be classed as having symptoms on the basis of a general question that doesn’t refer to specific symptoms. The section on long Covid, that asks about symptoms still persisting after at least 4 weeks, has the longer list of 21 symptoms, but that hasn’t been on the questionnaire for so long, and so the comparison with controls about this longer symptoms list can’t be made yet.

“More broadly, an issue is that all these symptoms are self-reported, rather than having been diagnosed and checked by health professionals. But I think there’s no good alternative – many of the people who report long-lasting symptoms may not even have had the opportunity to talk to a health profession about them.”



All our previous output on this subject can be seen at this weblink:



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.



in this section

filter RoundUps by year

search by tag