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

expert reaction to preprint giving the interim results from round 9 of the REACT-1 study on COVID-19 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A preprint, an unpublished non-peer reviewed study, reports on the latest data from the REACT-1 study on COVID-19 spread across England (swabs taken between 4 and 13 February 2021).

This Roundup accompanied an SMC Briefing.


Dr Tom Wingfield, Senior Clinical Lecturer and Honorary Consultant Physician, Liverpool School of Tropical Medicine, said:

“The REACT-1 Round 9 report on prevalence of SARS-CoV-2 prevalence in England in February 2021 has two key findings:

“1) The current national lockdown is having the desired impact of reducing SARS-CoV-2 prevalence: prevalence fell by two thirds since January (1.57% to 0.51%).

“2)  SARS-CoV-2 prevalence remains high: prevalence is similar to that seen in late September 2020 and, importantly, highest in people of Asian ethnicity, and people living in larger households and/or poorer neighbourhoods.

“While there are most certainly harsh socioeconomic consequences of lockdown, especially for children, the REACT-1 results can directly inform the precautions we should take when considering a roadmap towards lifting lockdown measures in England and the UK.

“In retrospect, the easing of lockdown restrictions after the first Covid-19 wave in 2020 was too early and took place at a time when the NHS Test and Trace system had limited coverage and was not functioning optimally.  We need to learn from this as we lift the current lockdown measures and not make the same mistakes again.

“Broadly speaking, reducing the prevalence of SARS-CoV-2 to as low a level as possible at the time of lifting lockdown measures will – alongside other measures such as suitable border control – provide the best environment for successful Covid-19 control and potential mitigation of future Covid-19 waves, including those related to new variants.

“This is hugely important when we consider that the REACT-1 study shows that people who currently have the highest prevalence of SARS-CoV-2 – people who are poorer, of Asian ethnicity, and live in larger households – are also those who are most vulnerable to severe Covid-19 disease.  This again lays bare the unacceptable socioeconomic and ethnic inequalities within our society.

“Finally, as a hospital doctor working in Liverpool, if lockdown is lifted at a time when SARS-CoV-2 prevalence is still relatively high, I am unsure how the NHS and its workforce – which is overstretched, understaffed, and exhausted – will be able to cope with future Covid-19 waves and continue to provide good care to people unwell and/or hospitalised with Covid-19 and other illnesses.”


Prof Lawrence Young, Professor of Molecular Oncology, Warwick Medical School, said:

“Latest data shows a significant fall in SARS-CoV-2 infections in England with reductions across all age groups in most regions.

“The overall decline in prevalence as determined in samples collected between 4th and 13th February was almost 70% as compared to the previous survey in January (weighted prevalence of 0.51% compared to 1.57% in samples taken between 6th to 22nd January 2021).  The largest falls were in London, the South East and West Midlands.  The fall in prevalence was similar in those aged 65 and over as compared to other age groups.  As almost all people over 70 years have had their first vaccine dose, the similarity in reduced prevalence trends in all age groups suggests that this is due to reduced social interactions during lockdown rather than vaccination.

“As with previous surveys, the study continues to highlight the higher infection levels in different communities associated with household size, living in a deprived neighbourhood and Asian ethnicity.  It also shows that healthcare and care home workers are more likely to test positive as compared to other workers.

“Overall levels of infection remain high and at a level similar to that seen in England in September 2020.  Along with the high number of COVID-19 cases currently in hospital, this data reinforces the need for caution in easing social distancing restrictions too quickly.”


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

“This new interim report from the REACT-1 survey, led from Imperial College, gives us some more good news – but there are cautions attached.  The report covers results from the first part of their Round 9, and is based on swabs taken from a reasonably representative sample of people aged 5 and over from the community population of England between 4 and 13 February.  That’s obviously very current.  Overall, REACT-1 are estimating that 0.51% of that population would test positive for the virus that causes Covid-19 – putting it another way, that’s about 1 person in every 195 testing positive.  There’s a bit of statistical uncertainty about that estimate, as is always the case with estimates based on surveys.  The researchers estimate that the proportion testing positive could be somewhere between 1 in 170 and 1 in 220.  But in any case it’s a huge reduction in infections from the last REACT-1 report, which covered 6-22 January.  According to the REACT-1 team, the number of people testing positive in 4-13 February is less than a third of number in 6-22 January.  That is excellent news, though the researchers do point out that the prevalence of infections is still around the level of late September last year.  There’s still quite a way to go before we are down to the relatively low levels from last summer, and we’ve seen how things could build up from that level to the second wave.  Because of this big decline in infections, the REACT-1 researchers estimate that the national R number for England is between 0.69 and 0.76, so almost certainly well below 1, and that the number of new infections is falling at between 4% and 5% a day.

“Of course, back last summer we didn’t have a vaccine, and now we have a large and rapidly increasing number of people who have been vaccinated.  We’ve heard a lot in the last couple of days on how there might already be evidence that vaccination is reducing rates of death, and possibly of serious illness, in the older age ranges where most people have now been vaccinated.  Interestingly, though, the REACT-1 researchers don’t attribute the fall in infection rates to vaccination.  They point out that the decline in positivity in the oldest age group that they consider, those 65 and over, is similar to the fall in the younger age groups.  Since it’s the over 65s, indeed the over 70s, where the biggest proportions of people have been vaccinated up to now, if vaccination were making the difference, you’d have expected the 65+ group to have shown much the biggest improvement, and that hasn’t happened.  The REACT-1 team believe that instead the observed falls are most likely due to reduced social interaction during lockdown, because that affects all ages.  I very much agree with that.  I’d hope that there might be some effect of vaccines in later REACT-1 rounds, and maybe even in the rest of the current round – but that’s not certain.  We know from the vaccine trials and from early data from Israel that vaccination should have a big effect on serious Covid-19 illness and death, but the evidence that vaccines reduce infectivity isn’t so strong yet.

“It’s not only the case that positivity reduced across all age groups in this study.  The REACT-1 researchers estimated the positivity for nearly 50 different subgroups of the population, divided up by gender, region, type of employment, ethnic group, household size, and more, and the infection rate fell substantially in all but one of those groups where they had sufficient data to make a proper comparison.  So good news overall, not just for some groups of people.

“As always, the REACT-1 report comments on differences in infection rates between different groups of people.  One effect of the greatly reduced level of infection, though, is that these comparisons become less easy to make.  That’s because, as the number of positive swabs in a group of people comes down, the statistical uncertainty about the positivity rate in the subgroup becomes greater.  But there’s still some evidence that infection rates are higher in people who work in health care or care homes, in people living in larger households, in people of Asian ethnicity (compared to White ethnicity) and in people living in more deprived areas.  These patterns may well become clearer after more data have been collected in the rest of this round – other clear comparisons of ethnicity were just not possible on the data so far, for example.

“There are also some quite substantial differences between regions of England in the REACT-1 results, with more infections being found in the Northern regions (North East, North West, Yorkshire and the Humber) than elsewhere.  That’s not a new pattern, but what is probably more important is that REACT-1 found that infections had decreased more slowly there than in the rest of the country.  The numbers had fallen, but only by between a quarter or a third, compared to the fall by more than two thirds across the country on average and by four fifths in some regions further south.  Because of these smaller falls in the North, the REACT-1 team are not nearly so confident that R is below 1 in those regions – indeed in the North East they estimate that it is more likely to be above 1 than below.

“It’s in these regional results that the REACT-1 results differ most in detail from the most recent results from the ONS Infection Survey.  Those ONS results don’t cover the same period as the REACT-1 report – 31 January to 6 February in contrast to REACT-1’s 4-13 February.  Overall ONS don’t show as fast a decline in positive tests than in REACT-1, but that might change when more ONS results come in later this week.  But the recent ONS results do show much clearer evidence of falls in the Northern regions than REACT-1 found, and in relative terms ONS were estimating lower infection rates in the three Northern regions that REACT-1 are.  That may have something to do with statistical uncertainty, because the estimates for single regions are less precise than for the whole country, simply because the number of people swabbed in a single region is lower.  But these regional differences are something to keep an eye on as more data emerge from REACT-1 and from the ONS survey.  We’re so lucky here to have both of these surveys, that provide relatively unbiased estimates of infections across England, so that we can see how they compare.  Most countries of the world don’t even have one such survey.”

Further information:

“The exceptional subgroup where the positivity rate went up rather than falling was in people who had had contact with a suspected but not confirmed Covid-19 case.  That’s a very small group and the estimates are therefore particularly uncertain, so it’s plausible that the rate didn’t really go up even in that group.  Because the number of positive tests was very low in a few other small subgroups, it was not possible to estimate the positivity rate in the latest data, so no comparison could be made with the earlier round.

“These REACT-1 results also provide comparisons (in Figures 4 and 5) between the REACT-1 estimates and the numbers of new confirmed cases from the Government dashboard at  They do match up reasonably well.  One wouldn’t expect the match to be perfect, because REACT-1, like the ONS survey, is measuring prevalence of infections (that is, the number of people who would test positive on a given day, regardless of whether they have a new infection or have already been infected for some time), while the confirmed case counts on the dashboard measure just new infections, that is, incidence.  These two measures can move differently at times when there’s a marked peak or trough in infections – but that’s not the case now, so it’s reasonably that they should match up, and match up they do, on the whole.

“One point about the mismatch between the ONS and the REACT-1 findings for the North East region is that the REACT-1 estimates of positivity there are really quite imprecise.  That’s because the number of people swabbed in the North East is relatively small.  In these new data, under 3,000 swab results were available for the North East, which is about half the number of the next smallest region, and there were over 19,000 swabs in the South East.  To some extent that’s not unreasonable, since the North East has by some margin the smallest population of all the English regions, but even allowing for that, the REACT-1 sample size is low in the North East.  It’s also low relative to the population size in the other two Northern regions, and very much so in London.  That’s a consequence of the way the REACT-1 sample is divided amongst different places – the aim is to get a big enough sample size in every local authority to allow a reasonable estimate of prevalence there to be made, and that means using a larger sample, relative to their population, in smaller local authorities than in larger ones.  And it happens that in the Northern regions and in London the populations of the local authorities are mostly relatively large, so they get rather less than what might be seen as their fair national share of the sample.  This imbalance is corrected in the weighting process that’s applied to the raw survey results, but it still does mean that the estimates for the North East, and to some extent for Yorkshire and the Humber, are relatively imprecise.”



Preprint (not a paper): ‘REACT-1 round 9 interim report: downward trend of SARS-CoV-2 in England in February 2021 but still at high prevalence’ by Steven Riley et al. was posted online at 00:01 UK time on Thursday 18 February 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