select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
A preprint, an unpublished non-peer reviewed study, reports on the latest data from the REACT-1 study on COVID-19 spread across England.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.
Dr Simon Clarke, Associate Professor in Cellular Microbiology at the University of Reading, said:
“As encouraging as it has been to see coronavirus infection rates falling for England as a whole, we still see high levels of infection across the country. The large drops in infection numbers seen in January and February have begun to slow, despite the increasing number of people who’ve been vaccinated. Worryingly, despite the vaccination programme, the numbers of infections have begun to go back up again in pockets of most English regions and the prevalence of infection in London and the South East is increasing.
“Data from previous REACT studies have corresponded well with daily hospital admissions, albeit with a lag of 16 to 22 days, dependent on age. Given the data in this report, we would normally expect to start to see the national average numbers of daily hospital admission begin to level off by next week (week commencing 8th March) and may even start to go back up in London and the South East, but that won’t happen if the vaccines have broken the link between infection and risk of hospitalisation. Moving forward, it will be fascinating to see whether this happens. Alternatively, if new virus variants that are either resistant to vaccine-induced or post infection immunity start to increase in numbers, or if immunity starts to drop away over time, we can expect to see a tighter correlation between community infection and hospital admissions.
“Given the slowing in the rate of decrease of infections, it remains essential that people who’ve been vaccinated continue to follow government guidelines so as to protect themselves and those close to them; it would be disappointing if those who have been the most vulnerable in our society put other people at risk, just because they’ve received their first dose a vaccine. The Department of Health might also reflect on their continued messaging surrounding the otherwise successful vaccine rollout. It could be a grave error to encourage people to underestimate their vulnerability to Covid-19.
“Data from REACT have consistently been reliable and high quality. This version uses the same hitherto successful and robust strategy.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This new report from the REACT-1 study covers the rest of their Round 9 of swab tests. The very brief summary includes good and bad news. On average, infections have continued to fall across England. But the speed at which they have fallen may have slowed somewhat. The pattern of change across the country isn’t uniform, and there’s evidence that infections may be increasing in several places. In a way, these differences between places are not surprising, since the pandemic has changed at different rates in different places right since it first began, but I can’t deny that it’s a bit discouraging to see some evidence of increases even as the overall trend continues downwards.
“As always, the study involves taking swabs from a reasonably representative sample of the community population of England aged 5 and over, and testing them for the presence of the virus that can cause Covid-19. Because the people tested are chosen to represent the national population, and not because they have symptoms or have been asked to have a test because of their job or some other reason (like a spike of infections in their area), REACT-1 can produce estimates that aren’t affected by the vagaries of symptoms and local conditions. The last REACT-1 report covered the first part of round 9, with swabs taken between 6 and 12 February, and the new report takes things forward to include the period 13-23 February.
“The main headline in the last report was that infection levels had fallen very considerably since the previous REACT-1 round in early to mid January. In that previous round, the researchers had estimated that 1.57% of the population would test positive for the virus – that’s about 1 in 65 of the population. In the first part of round 9 (6-12 February) that had fallen hugely, by about two thirds, to 0.51%, or about 1 in 195. In the latest data for 13-23 February, the national estimate has gone down again, but not all that much. The estimated percentage testing positive is now 0.47%, or about 1 in 210. That’s still a reasonably substantial fall in not much more than a week (on average) between the two halves of the round, but the REACT-1 researchers explain that that the rate of decrease per day is slower than it was between the January round and their last report before this one. The rate of testing positive has fallen in the second part of round 9, compared to the first part, in many, but not all, of the subgroups of the population that the report considers – but again, I think it’s not surprising and not necessarily worrying that the pattern isn’t exactly the same across all those groups. You wouldn’t expect absolutely everything to change in lockstep in such a short period. It’s worth mentioning that the currently estimated positivity rate from REACT-1 is lower than the most recent estimate for England from the ONS infection survey1, which also tests a representative sample from the community population. That ONS estimate was for 13-19 February, and it was 0.69% or 1 in 145. The new REACT-1 report covers a slightly longer period of time, up to 23 February – but we’ll see, when the next ONS infection survey report comes out later this week, how the difference between them might have changed.
“REACT-1 provides estimates of the R number and the daily rate of growth or decrease of infections. Just based on the decrease in infections between the two parts of round 9, the researchers estimate that R is between 0.76 and 0.97, which is not too different from the Government’s latest range for R (0.7 to 0.9) – but the REACT-1 researchers are confident that, nationally, R was below 1. The estimated growth rate from those two part-rounds of REACT-1 is between -4.1% and -0.4% per day, which is again not too different from the Government’s latest range of -5% to -3%, and still clearly indicates decreasing infections (because of the minus signs). Overall, the estimates from REACT-1 are slightly more pessimistic than the latest Government estimates, but they are still clearly indicating that infections are decreasing nationally, on average.
“The slightly more worrying news from this REACT-1 report comes from looking at data for the regions and smaller areas of England. Between the January REACT-1 round (round 8) and the whole of Round 9, the researchers found substantial decreases in infections in seven of the nine English regions, with smaller falls (but still falls) in the North East and in Yorkshire and the Humber. But comparing the two parts of the February round (round 9), there did appear to be falls in four regions (North East, North West, East of England and South West), no apparent change in Yorkshire and the Humber, and some evidence of rises in the remaining four regions (London, South East, East Midlands and West Midlands). It should be said that this evidence is not entirely clear in statistical terms. The confidence intervals, that indicate the statistical uncertainty in the estimates, overlap between the two halves of the round, so it’s not at all certain that the rates really have increased in those regions. But the REACT-1 researchers think it considerably more likely than not that infection rates did increase, over the course of the latest round, in London and in the South East. Again, that doesn’t match the conclusions in the most recent report from the ONS Infection Survey, which said that the percentage of people testing positive was falling in all regions with the possible exception of Yorkshire and the Humber (where the trend was uncertain). You must bear in mind, though, that the regional figures (from both surveys) are considerably less precise, statistically, than the national estimates, simply because the numbers of people swabbed in a single region is much less than the national number.
“On the same theme of differences between different places in infection trends, though, the REACT-1 researchers have carried out an analysis that indicates that there are many smaller areas of England where infections appear to have increased in the second part of round 9, compared to the first part. (This is shown in the third frame of Figure 6 in their report.) The areas of possible increase don’t fit clearly into the standard regions of the country, though they do broadly divide the country into fairly extensive areas where infections may be increasing, and other (on the whole larger) areas where they may be decreasing. What’s interesting and possibly important for these analyses is that most of the standard regions include extensive areas where infections appear to be increasing as well as areas where they appear to be decreasing – so that looking simply at what’s happening at the level of whole regions may be missing important detail. There are some particularly marked differences in London (Figure 7 in the report), where decreases in the north-east of the city are alongside quite substantial increases in parts of the west and south. I should caution, though, that these conclusions are based on quite small samples of swabs within these smaller sub-regional areas, and at changes over a short period between the two parts of the latest round. So I’d say they are signs for caution, not for alarm.
“There are a couple of other interesting new features in this report, that I’ll mention briefly. First, the researchers have supplemented their usual comparison of rates between different job types and different ethnic groups, by analyses comparing more detailed subgroups of the population. That’s interesting, but again the relatively small number of people in the sample in many of the subgroups manes that the estimates cannot be very precise. And there is a comparison between the trends in levels of infection (swab positivity) estimated in the survey and the daily numbers of hospital admissions. Those trends do align really quite closely, though (of course) there is a lag between the data on infections and the hospital admissions, since it takes time after a person is infected before they may become ill enough to need to go into hospital (with an estimated 18 days, on average, between these events).”
Preprint (not a paper): ‘REACT-1 round 9 final report: Continued but slowing decline of prevalence of SARS-CoV-2 during national lockdown in England in February 2021’ by Steven Riley et al. was posted online at 00:01 UK time on Thursday 4 March 2021. This work is not peer-reviewed.
All our previous output on this subject can be seen at this weblink:
www.sciencemediacentre.org/tag/covid-19
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.