select search filters
briefings
roundups & rapid reactions
factsheets & briefing notes
before the headlines
Fiona fox's blog

expert reaction to preprint giving the interim results from round 7 of the REACT-1 study on COVID-19 spread across 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A preprint, an unpublished non-peer reviewed study, reports on the latest data from the REACT-1 study on COVID-19 spread across England.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 Mark Woolhouse, Professor of Infectious Disease Epidemiology, University of Edinburgh, said:

“The REACT-1 survey continues to provide invaluable data on the state of the COVID-19 epidemic in England.  The survey provides estimates of the proportion of people with a detectable infection, whether or not they have symptoms.  Currently this proportion is about 1 in 150 people.

“In the most recent survey approximately 40% of those who tested positive did not have symptoms at the time the sample was taken.  This is about 1 in 200 people, which equates to over 250,000 cases across England.  These results support to the government’s intention to make mass testing as widely available as possible so as to find large numbers of people who may be infectious without knowing it.

“The survey also tells us about trends in prevalence.  The authors report a fall of around 30% between the last two surveys, confirming that cases in England are now falling.

“It is worth noting that, according to the REACT report, prevalence peaked one to two weeks before the England-wide lockdown began on November 4th.  This suggests that the measures taken prior to lockdown did reduce the R number to around one or possibly below one, though it is likely that R has fallen further since lockdown began.

“One exception to these trends is teenage school-children.  Prevalence continues to increase in this group and stands at about 1 in 50.  The report does not state how many of these were asymptomatic infections but it remains true that it is extremely rare for this age group to suffer serious illness.”

 

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

“The latest REACT-1 survey brings welcome news of a reduction in infections between 13 to 24 November: from 132 to 96 per 100,000, representing a 30% reduction (compared to the period 26 October to 2 November).  Across England, cases are halving every 37 days, reflected in an R number of 0.88 meaning that on average, every 100 infected people are passing on the virus to 88 other people.  But there is mixed picture across the country; infections are falling more quickly in the North, but not so quickly in the Midlands, South or London.  This probably reflects the fact that large parts of the North were under more severe restrictions before going into lockdown and should be a warning to MPs and councils lobbying for their local areas to be placed in lower tiers.”

 

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

“There’s some good news in these interim results from the first part of Round 7 of Imperial College’s REACT-1 infection survey, though there are reasons to be cautious as well.  The results, based on swab tests done between 13 and 24 November, show a fall of about 30% in the number of people in the English population who would test positive for SARS-CoV-2, compared to the second half of REACT-1’s round 6, done between 26 October and 2 November.  To be a little more precise, REACT-1 estimated that about 132 people per 10,000 would test positive in the second half of round 6, and about per 10,000 in the first part of round 7 – and 96 is about 30% smaller than 132.  A fall in the number of people testing positive is very broadly in line what the ONS statisticians reported about the latest estimates from their infection survey1, late last week, though ONS were rather more cautious in their conclusion and said that they believed the recent increase in infections was “levelling” rather than that infections had begun to decrease.  The most recent data from Test and Trace, and the data for new confirmed cases on the dashboard at coronavirus.data.gov.uk, also showed falling numbers of positive infections, though surveys of representative population samples like REACT-1 and the ONS Infection Survey are likely to produce much more reliable views of the trends.

“We’re very lucky in England to have two different regular surveys measuring infections.  I don’t know of any other country that runs even one regularly repeated survey like these.  (The ONS survey also runs in the other UK countries, but REACT-1 does not.)  The two surveys do not give exactly the same results for the positivity rates.  That’s not surprising – they draw their representative samples from different sources, some of the procedures are different, and they time periods about which they report are not always the same.  But, on the whole, they do report similar things about overall changes in infection patterns across England.  As well as the overall reduction in positivity rates across England, as usually the REACT-1 report gives estimates of positivity for almost 50 different subgroups of the population, defined by gender, age, the region where they live, ethnicity, and a lot of other characteristics.  The percentage of positive tests fell for the great majority of those subgroups between the second half of Round 6 and the first part of Round 7, though by different amounts for different subgroups, and there were increases in a few (for example, for younger people aged  between 5 and 17, and for people of Asian and Black ethnicity).

“Overall, REACT-1 estimates that the positivity rate in England, for the period between 13 and 24 November, was 96 per 10,000 people.  The most recent estimate from the ONS infection survey covers slightly different dates, 15 to 21 November, and is 116 per 10,000 people, so rather higher.  But both of these numbers come from samples, and so are subject to statistical uncertainty.  For REACT-1, they report that a plausible range for the true figure goes from 87 to 105 per 10,000, and the corresponding range for the ONS estimate is from 110 to 123 per 10,000.  These ranges don’t quite overlap, but given that the time periods are not quite the same, and given that ranges like these never capture all the uncertainty in the figures, they are roughly saying the same thing.  And a possibly important point is that, in almost all previous rounds of REACT-1, their estimate of the proportion who would test positive has been rather higher than the ONS estimate for roughly the same period, but this time the REACT-1 figure is lower than the ONS figure.  (There’s no clear way of saying which is more likely to be correct.)

“It’s important not to get too excited about these figures.  It certainly is good news that the infection rate, that has been increasing since the start of September, is showing some very clear signs of falling again.  But it hasn’t fallen very far yet.  According to both REACT-1 and the ONS survey, the infection rate in the most recent data is roughly the same as it was in mid to late October.  In very round numbers, based on the REACT-1 findings, that’s about double the rate at the end of September, and roughly 8 times as big as the rate at the start of September.  You would have to go well back into the first wave of infection to find positivity rates of around 1% (or 100 per 10,000), as REACT-1 are now reporting.  Neither REACT-1 nor the ONS survey were yet in operation during the first peak of the pandemic – they both started around the beginning of May, and already by that time both were estimating positivity rates of roughly 20 per 10,000, or very roughly one fifth of what they are now.  Things have started moving in the right direction again, but we’re by no means in the position we were at the end of the summer, or even the start of the summer, and that’s after a new lockdown and some fairly severe restrictions before that.  We can’t stop taking great care yet by any means.

“The regional positivity rates from this REACT-1 report do, very broadly, match the latest ONS estimates.  Between the second half of round 6 and the first part of round 7 on REACT-1, infection rates fell fastest in the regions of the North of England where they were highest to begin with, and fell less fast in the Midlands and South.  The upshot of these changes is that, on the latest REACT-1 estimates, infection levels appear to be higher in the East and West Midlands than they are in three Northern regions (North East, North West, Yorkshire and the Humber).  The largest differences between the regional positivity rates estimated by REACT-1 and the ONS survey are for the three Northern regions, where the REACT-1 estimates are lower than the ONS estimates – though again this could be because of small differences in the dates and in the detailed survey procedures.  For the other regions, the differences between REACT-1 and the ONS survey are smaller and not big enough to be outside statistical margins of error.

“It’s relevant that this is only an interim report on Round 7.  When the REACT-1 team have produced interim reports part-way through a round before, the details of the final report on the round didn’t match the interim findings in every respect, so we should in any case be a little cautious about the interim conclusions.

“One point about the press release is that I believe one line in it is potentially misleading.  It says, “Interim findings show a 30% fall in infections nationally between 13th and 24th November, with 96 people per 10,000 infected.”  This appears to refer to a change in infections over the very short period from 13 to 24 November.  In fact, there is no mention in the text of the preprint report giving a figure for the change over this period.  There are several mentions of a fall of about 30%, but they refer to the change between the second half of Round 6 (26 October to 2 November) and the first part of round 7 (13-24 November).  It’s not impossible, though, that the sentence in the press release is indeed referring to the period from 13 to 24 November, because Figures 1 and 2 in the preprint do appear to indicate a fall of about that size between those dates.  But the estimates for the infection rate changes in single days over this very short period are, statistically, subject to very large uncertainty, and I wouldn’t regard such a statement about that short-term change as being sufficiently reliable.  I suspect (though of course I don’t know for sure) that the authors of the preprint probably took the same view, otherwise they would have drawn attention to this difference in the preprint text.”

1 https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/coronaviruscovid19infectionsurveypilot/latest

 

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

“REACT-1 has released its exciting post-lockdown report for interim 7th round of swab-testing (75% sensitivity assumed).  In total, 105,000 persons aged 5+ years participated during 13th to 24th November 2020 (round 7a), over a week after England’s second lock-down which came into effect.

“Welcome message: Round 7a’s overall weighted prevalence of SARS-CoV-2 infection in the community in England was 0.96% (0.87% to 1.05%), that is: 9.6 per 1000 persons (down from 13.0 per 1000 during 16 October to 2 November).

“During the 13th to 24th November, R was estimated at 0.71 (0.54 to 0.91) with swab-test-positive infections halving every 5-weeks, that is: 37 days (95% uncertainty: 30 days to 47 days).

“Reason for weighting: The most deprived quintile contributed only 9% (9,500 nearest 10) of REACT-1’s 105,000 participants in round 7a – not the 20% that should be expected.  Re-weighted prevalence was 1.5% (1.2% to 1.9%) in the most deprived quintile vs 0.67% (0.56% to 0.8%) in the least deprived quintile (30,500 participants).

“North West versus North East: Based on rounds 6b and 7a, REACT-1 estimated the two regions’ reproduction rate as closely similar: for NW at R = 0.76 (95% CI: 0.71 to 0.82) and for NE at R = 0.76 (95% CI: 0.67 to 0.90).”

 

 

Preprint (not a paper): ‘REACT-1 round 7 interim report: fall in prevalence of swab-positivity in England during national lockdown’ by Steven Riley et al. was posted online at 00:01 UK time on Monda7 30 November 2020.  This work is not peer-reviewed.

 

 

Declared interests

Prof Mark Woolhouse: “No CoIs to declare.”

Prof Kevin McConway: “I am a Trustee of the SMC and a member of the Advisory Committee, but my quote above is in my capacity as a professional statistician.”

Prof Sheila Bird: “SMB is a member of the Royal Statistical Society’s COVID-19 Taskforce.”

None others received.

in this section

filter RoundUps by year

search by tag