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

expert reaction to preprint with results from round 15 of the REACT-1 study on SARS-CoV-2 prevalence in England

A preprint, an unpublished non-peer reviewed study, reports on the latest data from the REACT-1 study on COVID-19 spread across England.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

“This preprint is a report on the whole of Round 15 of the REACT-1 survey of Covid infections in England. That round is based on PCR swab tests and questionnaires from a reasonably representative sample of the English community population (aged 5 and over), collected between 19 October and 5 November. A previous interim preprint report, covering just the first part of Round 15, was published a couple of weeks ago, but this full report contains more detail and its estimates are generally more accurate statistically, because they are based on more samples.

“Several of the points covered in this report have already been covered in other data sources, including the other major infection survey in the UK (the ONS Coronavirus Infection Survey or CIS), and/or in the REACT-1 interim round 15 report – but that is in itself, arguably, a reason why REACT-1 is useful and important. The two surveys (REACT-1 and the CIS) avoid the biases that can arise in routine counts of confirmed cases (such as on the dashboard at because of changes in the numbers and types of people coming forward for testing. Instead the surveys test people regardless of any symptoms they might have, simply to plot the track of the pandemic. In my view, it’s very useful to have both of the surveys, because they have differences in the details of how they choose the people to be tested and how they analyse the data, and one survey might turn out to have quirks that make its results less reliable. With two of them, there is the possibility of comparing the findings to check that they broadly agree, which, on the whole, they do. There are also opportunities to estimate details in one survey that were not investigated in the other one.

“REACT-1 round 15 shows a higher estimate of the percentage of people in England that would test positive for the virus, than was found in round 14 that was carried out during September. That’s no surprise at all – every estimate of infections or cases has shown a substantial increase since September in England. For round 15, REACT-1 estimates that , on average, 1.57% of the English community population would have tested positive at any time between 19 October and 5 November. That’s about 1 in every 64 people, and it is the highest positivity rate ever estimated in any REACT-1 round. (The estimate for round 8 in January this year was exactly the same, but no estimate has been higher.) The average positivity estimate from the ONS CIS, for those dates, is actually a bit higher than that, at 1.93% or about 1 in every 52 people, but the difference between the two estimates isn’t immense and does not concern me, given differences in methods and also in the ages of the children included. (CIS goes down to age 2 years.) The short-term trends during the period of Round 15 look reasonably similar between the two surveys – in their latest reports, both are saying that positivity increased overall until somewhere roughly around October 20, and then fell back a bit (though it remains high).

“Some of the specific details of positivity rates in subgroups, such as age groups or different regions of the country, do differ somewhat between REACT-1 and the CIS, but that’s not really surprising – statistically, the estimates for subgroups are bound to be less precise that for the whole English population, just because fewer people are swabbed for the survey in a subgroup than in the country as a whole. But the main general features do generally match across the surveys.

“As in all the other data sources, this REACT-1 round found that infection rates were highest in children of school age. The estimated positivity rate in children aged 5 to 12 was 4.95%, or about 1 in 20 testing positive. For the 13-17 age group the estimated rate was slightly higher at 5.21%, or about 1 in 19. The positive news on that is that the REACT-1 researchers estimate that the positivity rate in those age groups fell quite substantially during the period of the round, and indeed there was also a fall (though not so great, proportionally) in the positivity rate in those aged 18-54. The positivity rate in those aged 55 and over did not, however, show any evidence of falling during the round, and may have increased slightly. The positivity is, however, much lower for those older age groups than for schoolchildren (at about 1 in every 115 for ages 55-64, 1 in 120 in ages 65-74, and about 1 in 160 for people over 75 and over. Of course, there can be no guarantee that the positivity rate in children will continue to decrease – this round of REACT-1 finished about a week after the half-term holiday in most of England, and many infections detected at the end of the round would have been incurred a bit earlier than that (because of the brief incubation period between a person being infected and the virus becoming detectable), perhaps during half-term. After half-term the dynamics might change.

“There is interesting new data on vaccine effectiveness in this preprint. The researchers make an important point about measuring vaccine effectiveness in surveys such as this. For many reasons, there will be differences, possibly quite large ones, between the group of people who are vaccinated and the group who are not vaccinated, apart from the difference in vaccination status. So any difference in the rate of testing positive in the two groups could, in whole or in part, be caused by those other differences and not the vaccines at all. Statistical adjustments can be made to try to allow for these other differences, and that was done in this study, but those adjustments can only go so far. Since, particularly in older age groups in England, the vast majority of people have been vaccinated, the differences between those who have been vaccinated and those who remain unvaccinated are quite marked. Also the estimates of positivity in the unvaccinated group will not be very precise, statistically, because the number of unvaccinated people in older age groups is so small nowadays. So, rightly in my view, the researchers do not present detailed overall estimates of vaccine effectiveness comparing all vaccinated and all unvaccinated people.

“They do argue, though, and I agree, that estimates of the effectiveness of a booster dose of vaccine won’t be subject to the same kind of biases, because there will be fewer other differences between people who have had two doses and people who have had three, and because the two-jab and three-jab groups are closer in numbers. They also argue that they have seen fewer differences between vaccinated and unvaccinated children aged 12-17 than in older age groups, and also the numbers of vaccinated and unvaccinated children are again closer than for older people. So the researchers do publish estimates of vaccine effectiveness against infection for third doses in adults, and for simply being vaccinated (compared with not vaccinated) in children aged 12-17.

“The findings on vaccine effectiveness for the third dose are that, having made the appropriate statistical adjustments, people who have had the third dose have a chance of being infected that is about 38% of the chance of infection of people who have had only two doses. That means that the estimated vaccine effectiveness of the booster dose, compared to having had two doses but no booster, is 62%. That is a very substantial effectiveness, given that it comes on top of the effectiveness of two doses.

“The estimate of vaccine effectiveness in children aged 12-17, comparing those who have had at least one dose with those who have not been vaccinated at all, is about 56%. That’s the effectiveness against testing positive for an infection. The estimated effectiveness against symptomatic infection is higher, at about 69%.”


Prof Paul Hunter, Professor in Medicine, The Norwich School of Medicine, University of East Anglia, said:

“There are now very many different sub-lineages of delta which started off as B.1.615.2 though sub lineages are designated AYx.x such as AY.4.2

“Both AY.4 and B.1.615.2 were detected in the UK last year according to COG-UK data. They have been circulating in the UK since the end of last year though at very low numbers until Easter time. From May of this year AY.4 has been the dominant delta variant. But it doesn’t appear to be any different in its infectivity to the other delta variants other than AY.4.2.

“AY4.2 first appeared around May June of this year and has been gradually outcompeting the other delta variants though not that rapidly and it is still present in only a minority of infections (<20%). At present it seems to be about 10% more infectious than the other delta variants, though I am not sure whether we know why this is yet. Last time I asked a viral geneticist he wasn’t sure so I am not going to speculate.

“The results of REACT round 15 on distribution of variants is consistent with that of the national screening system as presented on COG-UK.

“I think it is too early to say for certain on this data whether AY.4.2 is really less likely to cause illness than other variants. If this association is confirmed then there are several possibilities, one of which is the variant is indeed less virulent or that it is more likely to cause infections in people who are partially immune. More work is needed. It would be good, however, if this variant was indeed less virulent but it would be wrong to give too much credibility to this observation just yet.”


Dr Simon Clarke, Associate Professor in Cellular Microbiology, University of Reading, said:

“This latest REACT study paper covering 19th October to 5th November shows a mixed picture of infections where an initially increasing number of infections peaked around the 20th/21st October, then began to decrease.  This reflects what was seen in the government’s national coronavirus data, which relies on people getting tested when they experience symptoms.  REACT is a more random way of sampling across the population, so we can be confident that the decreases seen in the government’s numbers were indeed representative.

“The government’s data have recently shown an increase in infections, highlighting how dissection of the daily ups and downs in the number of infected individuals does not offer us any guide to what is going to happen beyond the immediate term.  Infection numbers are a situation report, not a projection of what is going to happen in the coming weeks.  The future remains uncertain.

“Interestingly, REACT observed that the AY.4 variant accounted for more than half of the isolates which had their genetic code sequenced, showing how much it has spread in England.  The AY.4.2 variant was reported and shown be slightly less likely to cause symptomatic infections that AY.4. The data do not indicate the severity of those symptoms or who they occur in. If these AY.4.2. infections occurred in younger people or in communities where vaccine uptake is relatively high, these factors could account for the observed difference. Similarly, if AY.4.2 is a more transmissible variant able to infect more people, that alone could cause a greater burden of human disease, regardless of any possible decrease in ability to cause disease.

“It should not be forgotten that we have seen multiple evolutionary steps where the virus has become increasingly able to cause disease.  If AY.4.2 is indeed less able to cause disease than A.Y.4, the coronavirus will likely still be move virulent than the variants that caused the UK lockdowns.

“What is more certain from these REACT data is that vaccinating over-12s with a single dose reduces their chances of contracting virus and going on to act as a vector spreading it to friends and family.  No matter how effective the vaccines may or may not be at reducing coronavirus spread from an infected individual, if you don’t catch it in the first place, you can’t spread it.

“Alarmingly, nationally about 1 in 20 school age children were infected with the coronavirus over this stage of the REACT study.  This picture might improve with better vaccine rollout and uptake and if a second dose is eventually authorised for this age group.

“REACT continues to demonstrate vaccine effectiveness in adults. These latest data showed a marked difference in the risk of infection if they’d received a vaccine booster dose.  As we move into the winter months this highlights the importance of the government’s booster campaign in keeping a firm lid on both infection numbers and the risk of Christmas restrictions.”



The preprint ‘REACT-1 round 15 final report: Increased breakthrough SARS-CoV-2 infections among adults who had received two doses of vaccine, but booster doses and first doses in children are providing important protection’ by Marc Chadeau-Hyam et al. was posted online at 00:01 UK time Thursday 18 November. This work is not peer-reviewed.



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