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

expert reaction to latest results from the REACT-1 study on COVID-19 prevalence across England – round 16

A preprint, an unpublished non-peer reviewed study, reports on the latest data from the REACT-1 study on COVID-19 spread across England.

This Roundup accompanied an SMC Briefing.


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

“At these times of very high and rising infection rates, and of large changes over time in the numbers and type of people who are coming forward to be tested, because of concern about the rise of Omicron, Government advice to test before social events or large gatherings, concerns about the possibility of infecting vulnerable relatives and friends over Christmas, and so on, it’s really important to look at the findings from a survey like REACT-1. It tests a representative sample of the community population (aged 5+) of England, not because of any of the reasons I’ve just given for changes in the people being tested generally, but only to track the progress of the epidemic. So it shouldn’t be affected by biases that can arise with, for instance, the confirmed cases numbers on the dashboard at because of changes in who is being routinely tested.

“Concerns about those biases are greater than usual because of major rises in the number of tests being done – and they may get more severe, because the daily number of tests being done is beginning to approach the available testing capacity. If people can’t all get tested because of not enough tests are available, there will be yet more, and more complicated, biases in the dashboard confirmed case figures.

“But none of those specific concerns about bias apply to the tests carried out in REACT-1, or indeed in the other major infection survey in the UK, the ONS Coronavirus Infection Survey (CIS) which also tests a representative sample of people (of a rather different nature). It might seem like overkill to have two of these surveys in England, but in my view it’s an admirable thing. It’s true that, in broad terms, the findings from the two surveys tend to agree, but that means that each of them provides a check on the other. No survey is perfect, but having two of them that use rather different samples of people and rather different statistical methods allows us to see that nothing untoward is happening. We all know about ‘rogue’ opinion polls at the time of elections – they can be identified because there isn’t just one polling company and the results can be cross-checked, and it’s the same with the two infection surveys. Also each survey provides some results that the other does not.

“That said, most of the main findings in this REACT-1 round do line up with what we’ve seen in the ONS CIS and, to some extent, in changes in confirmed case numbers and in other data on the new Omicron variant. REACT-1 Round 16 took swabs between 23 November and 14 December, though the results based on genome sequencing (that identified the variant involved in positive test results) go up only to 11 December. As always, there was a gap between Round 16 and the previous round, that ended on 5 November, so REACT-1 can’t tell us anything about what happened between the two rounds.

“Overall, the prevalence of positive tests, as estimated by REACT-1, fell slightly between Round 15 and Round 16. In Round 16, the team estimate that, on average, 1.41% of the community population aged 5+ would have tested positive – that’s about 1 in 70. The corresponding figure for Round 15 was about 1 in 65, so not a huge change. But looking at an average across the whole of Round 16 doesn’t cover the detail of what went on during that round. The researchers estimate that there wasn’t much change in the proportion of the population who would test positive over roughly the first week of the round, which was the last week in November, but after that there’s clear evidence of an increasing trend in infections. Of course, that last two weeks or so of this REACT-1 round, roughly the first two weeks of December, cover the beginning of the period after the Omicron variant arrived in the UK, and the rise in infections in those weeks must have a lot to do with the rise of infections from that variant.

“Because REACT-1 aims to sequence all the virus genomes found in their positive tests, there is some information here directly on the rise of Omicron. The sequencing cannot always provide a clear definition of which variant is responsible for each infection, because (for instance) there may be insufficient virus material in the swab sample to identify the variant, but a large proportion of the positive samples do detect which variant is involved. The snag here is that, despite its rapid rise, there were not all that many Omicron infections in England up to 11 December, the last date for which sequencing results could identify variants vor this research report. The REACT-1 team found their first Omicron variant on a swab taken on 3 December. They report that there were 275 positive samples where the variant could be identified between 1 and 11 December, of which just 11 were Omicron, the rest being Delta. The report gives details of what sort of people were infected with Omicron, but honestly, with only 11 cases, no general conclusions could possibly be drawn from that. All but one of the Omicron cases were in London and the South of England.

“The researchers found differences in the way that infection levels changed in different regions during this round of the survey. The clearest features are a very marked rise in infection levels during the round in London, and also to a lesser extent in the South East, particularly later in the round. Again this matches what has been seen in the ONS CIS and in the dashboard case counts – and it does correspond to where most of the 11 Omicron cases were found, though I still don’t want to read much into just 11 Omicron cases, so far, in this survey.

“Other major findings relate to the level of testing positive in different age groups. The REACT-1 researchers observed a considerable fall in the prevalence of testing positive in older school-age children (aged 12-17), where a considerable amount of vaccination has been going on, but no change in prevalence in younger children (aged 5 to 11) who are, so far, mainly ineligible for vaccines. The prevalence of testing positive is the highest, out of all the age groups they consider, for the 5-11 group, with an estimated proportion of about 1 in 20 testing positive on average during Round 16. The prevalence is second highest in the 12-17s, with about 1 in 45 testing positive – but that is about half the rate of testing positive that they observed in that age group in Round 15 (19 October to 5 November), while the rate in the 5-11 group had barely changed since Round 15.

“The researchers also report that prevalence fell considerably in Round 16 compared to Round 15 in the two oldest age groups that they consider (65-74 and 75+). That’s particularly the case for the 75+ group, where the estimated proportion who would test positive fell from about 1 in 160 in Round 15 to about 1 in every 475 in Round 16. The researchers relate this to the booster vaccinations that most of the over 65s have had. I’m sure that’s relevant, but there may be other factors involved, such as some older people being particularly careful about social contacts or wearing masks.

“The REACT-1 researchers provide information on two aspects of vaccine effectiveness. You need to be a bit careful in interpreting those results. They are essentially arrived at by comparing the rate of testing positive in people who have had an aspect of vaccination with those who have not. There are likely to be other differences between those two groups apart from their vaccine status, and maybe those other differences would explain some of any difference in infection prevalence. However, the researchers look at only two aspects, and they make statistical adjustments to allow for some differences in other factors (survey round, age, sex, deprivation level where they live, region where they live, ethnicity). I suspect that this takes care of the important factors of differences (potential confounders) in the two aspects that they consider, though one can never be sure.

“What they found was that, in children aged 12 to 17, being vaccinated led to a reduction in the chance of testing positive by roughly half, compared to unvaccinated children of the same age range. (This is based on data from Round 14 to 16, not just round 16.) That’s quite a substantial decrease in the chance of infection. Evidence from other sources on how likely an infected person is to infect someone else – transmission – isn’t encouraging, in that it seems that being vaccinated doesn’t change the chance of transmission much. But those results are about an infected person’s chance of passing it on. A person who doesn’t get infected can’t pass it on at all, and these findings on vaccination effectiveness in children show that vaccination would, roughly, halve the number of children who could possibly pass on the infection to others.

“In adults (18 and over), the REACT-1 researchers found that having three vaccine doses led to a reduction in the chance of testing positive by a little less than three quarters, compared to people who had only two doses. (This result is based just on Round 16 data.) That’s a very substantial reduction, but it’s broadly in line with other estimates of the effectiveness of a booster dose, and figures like this are behind the big push to get as many people boosted as possible. We mustn’t forget that Omicron hadn’t become dominant during this REACT-1 round (though it was present later in the round), and the effectiveness of boosters is thought to be less for Omicron than for Delta. But other investigations have found that there still is a pretty strong protective effect of booster vaccination against Omicron.”



‘Rapid increase in Omicron infections in England during December 2021: REACT-1 study’ by Paul Elliott et al. is an unpublished preprint



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.”

in this section

filter RoundUps by year

search by tag