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

expert reaction to preprint with results from round 14 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A preprint, an unpublished non-peer reviewed study, reports on the latest data from the REACT-1 study on COVID-19 spread across England.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.


Commenting specifically on the report in the preprint that one instance of the Delta with E484K mutation was detected (“All positive virus samples analysed correspond to the Delta variant or sub-lineages of Delta with one instance of the E484K escape mutation detected”):

Prof Francois Balloux, Professor of Computational Systems Biology and Director, UCL Genetics Institute, UCL, said:

“Nothing to worry about in particular.  Delta is at >99.99% prevalence in the UK.  Roughly 0.2% of Delta variants carry E484K.  The mutation has been acquired independently multiple times by the Delta variant, but if anything, it seems to reduce the fitness of the lineage, and there is no evidence that Delta+E484K is a successful combination.”


Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:

“REACT-1 round 13 took place during 24 June to 12 July (consent-rate: 11.7%); round 14 in 9-27 September 2021 when consent-rate (as prompt joiners) had increased slightly to 12.2% but was far below 30% in REACT-1’s first round.

“REACT-1 is a hugely important study in England and merits a higher participation-rate than it currently enjoys.  Re-weighting would then be less necessary and representativeness better assured.  Nonetheless, consistency between REACT-1’s successive rounds is exploited to an enviable extent in excellent analyses.

“Participants are asked to give secondary consent: this time for linkage to their NHS records.  The secondary consent-rate was 87.5% in round 14 but could be higher still if citizens were asked for more limited consent such as to retrieve COVID-vaccine type and dates.

“Key findings in round 14 are, firstly, exponential increase in SARS-CoV-2 infections in children (aged 5-17 years) but shrinking epidemic currently in adults.

“Secondly when combining rounds 13 & 14, fully adjusted vaccine effectiveness against infection relative to those unvaccinated was 45% (95% CI: 22% to 61%) for Oxford/AstraZeneca and 71% (95% CI: 57% to 81%) for PfizerBioNTech.

“Thirdly, weighted prevalence of swab-positivity was 0.55% (95% CI: 0.50% to 0.61%) for those who received their second dose of vaccine 3-6 months before their swab compared to a substantially lower percentage at 0.35% (95% CI: 0.31% to 0.40%) for those whose second dose was within 3 months of their swab.

“Round 15 may tell us more assuredly about an elapsed time of more than 6 months for those unboosted.”


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

“The latest preprint from the REACT-1 survey study of infections with the virus that can cause Covid-19 covers Round 14 of the survey.  Unlike the ONS Covid-19 Infection Survey (CIS), which runs continuously, REACT-1 takes place in separate rounds.  (Also, the CIS covers the whole of the UK, though it publishes results separately for the four UK countries, while REACT-1 covers only England.)  The virus-testing swabs for Round 14, over 100,000 of them, were mostly collected during the period 9-27 September – this is quite a time after Round 13, which ran from 24 June to 12 July, so REACT-1 can’t tell us anything about how infection levels changed between mid-July and the second week in September.

“In terms of general patterns and trends in infection levels, the latest REACT-1 findings do broadly match what we’ve seen from other sources including the ONS CIS.  Rates of testing positive for the virus were somewhat higher overall in September than they were back in July, were increasing (slowly) during some of September, but the rates of testing positive, and the direction of the trend in testing positive, was very different from one age group to another, with increases in children of school age and below but decreases, mostly, in older age groups than that.

“However, the estimates of the positivity rates, rather than the trends in positivity rates, from these REACT-1 results do differ quite a lot from the estimates from the ONS CIS.  For the whole population (aged 5 and over), for the whole of Round 14, the REACT-1 researchers estimate that 0.83% would test positive – about 1 in every 120.  The average of the daily CIS positivity estimates for the period covered by that round, for England, is 1.17% testing positive, or about 1 in 85, quite a bit higher than the REACT-1 estimate.  Both REACT-1 and the CIS found considerable differences between the positivity rates in different age groups, but again the results from the two surveys do differ from one another quite a lot.  For Round 14, REACT-1 estimated the positivity rate for children aged 13 to 17 as 2.55%, or about 1 in 40, while the average ONS estimate for those aged from school year 7 (11 or 12 years of age) to school year 11 (16 or 17 years of age) is 3.70%, or about 1 in 27.  That’s quite a big discrepancy, and I don’t think it can really be explained simply by the age groupings being a bit different in the two surveys.  The estimates for the next age group up, so 18 to 24 for REACT-1 and school year 12 to age 24 for the CIS, which cover almost the same ages, are 0.46% (1 in 217) for REACT-1 and 1.40% (1 in 71) for the CIS.  For other age groups, the differences are mostly rather less marked than this, but they do still exist.

“There are plenty of reasons why one wouldn’t expect the results to match up exactly – the two surveys use different lists for choosing their samples to test (basically GP lists of individuals for REACT-1, and standard lists of households for the CIS), have rather different testing procedures, and so on – but these differences are quite large and need investigation in my view.  The REACT-1 researchers do explain that there have been some changes in the details of their procedures for testing and for getting the swabs to the labs that do the actual tests, and I am no expert on those technical details – could they have led to differences between the two surveys on this scale?  It may be an issue that the age groups I’ve mentioned above are groups where the positivity rate was changing quite fast over the period of this REACT-1 round, and generally in opposite directions (increasing for the 13 to 17 ages group, and decreasing for 18 to 24), and that probably makes such discrepancies more likely.

“As in the previous round, these REACT-1 results include estimates of vaccine effectiveness, based both on what vaccines the surveyed people reported that they had had, and also on what is actually recorded in their medical records, where participants agreed to have those linked to their REACT-1 records.  Those results allow estimation of how effective the vaccines are against all infections, whether or not the person had symptoms or not, because REACT-1 detects all infections.  To some extent, again, those results fit in with what is known from other sources – vaccines are somewhat less effective against any infection than they are against infections that lead to symptoms, though the effectiveness is still reasonably high against all infections, at an estimated 63% across all vaccine types.  (That means that, other things being equal, the chance of getting infected after two jabs is about 37%, a bit over two-thirds, of the chance of getting infected for unvaccinated people.)  Estimated effectiveness against infection was estimated to be rather higher for people who had had the Pfizer jabs than those who had had AstraZeneca, but the level of statistical uncertainty in those estimates was rather high so that it’s not completely certain that they do differ in effectiveness, and perhaps the apparent difference has something to do with the different time periods when different vaccines were most commonly in use.

“An observation that’s slightly concerning, perhaps, is that the rate of being infected, in fully vaccinated people, appears to increase as the time since they had their second dose increases.  That could perhaps indicate that the effectiveness of the vaccine is reducing over time – but, as the REACT-1 researchers point out, these estimates involve averaging over different periods of time, and different states of the pandemic, and maybe they aren’t comparing like with like.  (For instance, in order to get enough data to make statistically reasonable estimates of effectiveness, the researchers had to use data from Round 13 (back in June and July) as well as Round 14 in September, and the state of the pandemic was rather different generally in those two rounds.)  Also, one has to remember that many people whose had their second jab some time ago will now be receiving a booster jab anyway.”



Preprint (not a paper): ‘REACT-1 study round 14: High and increasing prevalence of SARS-CoV-2 infection among school-aged children during September 2021 and vaccine effectiveness against infection in England’ by Marc Chadeau-Hyam et al. was posted online at 00:01 UK time on Thursday 14 October 2021. 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