select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The Office for National Statistics (ONS) have published the latest data looking at prevalence of self-reported long COVID after two doses of a COVID-19 vaccine, received before infection.
Dr Claire Steves, Senior Clinical Lecturer, Kings College London, said:
“It’s great to see this data, which shows a reduction in risk of Long covid symptoms at 12 weeks after infection. It backs up findings we published last year, using a different dataset. They didn’t find statistically different findings between vaccine types, and also did not present findings on booster vaccinations.
“It’s another reason for people to be vaccinated, as well as reducing risk of severe initial illness.
“However, this study reminds us that while substantially lessened, vaccination doesn’t completely get rid of the risk of long Covid entirely, so with numbers so high it’s still important that we search for treatment strategies and ensure people affected have the support they need.”
Dr James Doidge, Senior Statistician, Intensive Care National Audit & Research Centre (ICNARC); and Honorary Associate Professor, London School of Hygiene and Tropical Medicine, said:
“Because of its systematic data collection, the Coronavirus Infection Survey is one of the most valuable data sources in the world. With the potentially large impact of long COVID on population health and healthcare, understanding the impacts of vaccination on prevention of long COVID is also important. For young, healthy people who have very little risk of developing severe COVID-19, prevention of long COVID could be the crucial factor in tipping the scales of risks versus benefit of vaccination. Unfortunately, this analysis does not stratify by age and with an average age of 47 and a standard deviation of 11 years among the unvaccinated group, it appears that very few young people were included. We already know from previous research – from this same study – that young people are at much lower risk of developing long COVID [1].
“Putting questions about the relevance to young people to one side, there is one glaring problem with this analysis. The analysis matches people who were infected after vaccination to others who were unvaccinated at the time of the infection, and ensures comparable distributions of age, sex, White or non-White ethnicity, country and region, area deprivation quintile group and health status. However, there is one crucial factor that the analysis does not control for: COVID-19 variant. Table 1 reveals that most of the unvaccinated group were infected during the wild or Alpha-dominant periods whereas most of the vaccinated group were infected during the Delta wave. We already know from other research that Delta is associated with about a 31% lower odds of having symptoms persisting to 28 days [2]. This analysis indicates a 41% difference at 12 weeks so it seems likely that much of this difference is explained not by vaccination but by differences in the variant to which each group was exposed. The authors do acknowledge the difference in timing of infection as a limitation but there is no acknowledgement of the statistical implications of this and it is not mentioned at all in the highlights, where it would provide crucial context. To present these results as if they are entirely attributable to vaccination, or apply to everyone equally, would be misleading. On the other hand, there is a growing body of evidence from other sources (e.g. [3]) that vaccination does reduce the incidence of long COVID. The important, unanswered questions are by how much and for whom?”
[2] https://www.medrxiv.org/content/10.1101/2021.11.24.21266748v1
[3] https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(21)00460-6/fulltext
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“These new findings from the ONS Coronavirus Infection Survey (CIS) are interesting and important, but they can’t tell us everything we want to know. Since they come from the CIS, they are based on data from a representative sample of the population, which is a strength. They show that the odds of having self-reported long Covid symptoms, lasting at least 12 weeks after the initial infection, in people who were double-vaccinated at the time of infection, were about 40% less than the odds for unvaccinated people. A very similar relative reduction in odds of about 40% applied to the odds of having long Covid symptoms that limited the person’s ability to undertake daily activities. (These are reduction in odds, not in the probability of having long Covid, though in round numbers the reductions in the probabilities are also around 40%.)
“Putting it another way: in 1,000 people like those whose data were analysed and who weren’t vaccinated, about 150 would report long Covid lasting at least 12 weeks. In another group of 1,000 who were very similar but were double vaccinated before they were infected, about 90 would report long Covid lasting at least 12 weeks, so about 60 fewer. However, as with all numbers based on surveys, there’s some statistical uncertainty about that number. The number reporting long Covid in the 1,000 who were double vaccinated could plausibly be between about 80 and about 105, so between 45 and 70 fewer than in the 1,000 who were unvaccinated.
“ONS are careful to mention that we can’t deduce from these results that double vaccination causes a reduction in the chance of long Covid. That’s because this is an observational study. People weren’t allocated at random to be vaccinated or not – they did what they would have done anyway, regardless of the survey. Therefore there will be many differences between the vaccinated and the unvaccinated people, apart from the simple fact of vaccination. Any combination of these other differences could be the real cause of the differences in the chance of long Covid, in whole or in part, and not the vaccination at all.
“ONS did make their comparisons by looking at groups of vaccinated and unvaccinated people who matched in terms of age, sex, whether they had White ethnicity, which UK country or region they lived in, what was the level of deprivation in the area they lived, and the time from infection to the follow-up interview where they reported their long Covid. They also made statistical adjustments to take into account differences in these factors that still existed after the matching process. But that doesn’t remove the issue of cause and effect entirely. There will be other factors that differ between vaccinated and unvaccinated people apart from those used for the matching, for example in the type of work they do, and it’s possible that some of those factors might affect the chance of long Covid.
“However, just because we can’t be sure that vaccination causes a reduction in long Covid risk, that certainly hasn’t ruled out the possibility that it does cause a risk reduction. I think these results do provide quite a big measure of indicative evidence that double vaccination might well cause a reduction in the risk of long Covid, if one is unlucky enough to become infected after being vaccinated. But we can’t be sure of that, and we also can’t be sure that the size of any reduction is definitely the same as found in these results.
“It’s also important to note that the data behind these findings were collected up to the end of November 2021, so they are too early to say anything specifically about the Omicron variant, or indeed to measure the effect of boosters.
“ONS report that there was “no statistical evidence” of a difference in the reduction in odds of long Covid according to whether people had had an adenovirus vaccine (usually Oxford/AstraZeneca) or a messenger RNA vaccine (generally Pfizer/BioNTech or Moderna). Figure 1 in the ONS bulleting does appear to show quite a large difference between the two vaccine types, in terms of the dots that show the central estimates of the odds reduction. But, first, the lines on the diagram showing the margins of uncertainty are rather long, for both vaccine types, and the margins of uncertainty overlap a lot. That means that it’s possible that the apparent difference between the vaccine types is entirely due to chance, in particular on who, by chance, was invited to take part in the survey. But also, there are differences between the people who took the two vaccine types, other than which vaccine type they had. For instance, younger people would be less likely to have had Oxford/AZ, because it was not recommended for them. Differences in age would have been accounted for, to some extent at least, by the statistical matching and adjustment for age. But younger people who were vaccinated early, because they worked in health or care or because they were particularly clinically vulnerable, might have had Oxford/AZ anyway, and the statistical analysis took no account of what jobs people did or their clinical status. So even if there really is a difference between the effect of the different vaccine types on long Covid risk (and there may not be a real difference), it isn’t necessarily caused by the vaccine type rather than something else. ”
Dr David Strain, Chair of BMA Board of Science and Clinical senior lecturer and honorary consultant. University of Exeter Medical School, said:
“These data are in keeping with the growing understanding that long COVID is, at least in part, immune mediated. Yesterday, we saw data that low levels of one of the immunoglobulins is associated with the risk of long COVID. These are stimulated by vaccination, therefore the evidence that vaccines are reducing the risk of progressing from covid to long COVID is to be expected. It is important to remember that this is in addition to the protection afforded against the initial infection – you can’t develop long COVID if you don’t get COVID in the first place.”
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.