select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
The government have published the latest estimates* for the COVID-19 R value in England and Wales, and the Office for National Statistics (ONS) have published** the latest data from their COVID-19 infection survey.
Commenting on the ONS Infection Survey data**:
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“The ONS infection survey continues to provide valuable evidence about the course of the pandemic, because it collects, analyses and publishes test results from a representative sample of the community population, originally just of England, but now of Wales too. (The survey has recently been extended to Northern Ireland, but has not yet been running long enough there to provide any publishable results.) They exclude people living in communal establishments like care homes or prisons, but the vast majority of the population is represented. The people in the survey are tested because they are in the survey, and not because of any COVID-19 symptoms they may or may not have, or because of their job, or because they have been traced as contacts of an infected person. Of course, they might happen to come into one of those other categories too, but that’s not why they are in the survey, and so these results are much less affected by the sorts of bias that can affect other data sources based on test results.
“For clarity, because there’s sometimes been some confusion about this lately, I should explain that the survey results provide three different basic sets of figures. First, there are estimates of how many people in the whole country would give a positive swab test, if they were tested to see if they currently were infected with the new coronavirus. ONS report, as they did last week, that those figures, for England, fell from when the infection survey started in April until the end of June, that after that there was a fairly small increase for a few weeks, but that that increase seems to have levelled off. Because all these results are based on a sample of the whole population, there is some statistical uncertainty in them, but the general trend over the past few weeks does look level, and there’s certainly no sign of a clear increase or decrease at present. The ONS release provides estimates for the English regions as well, but, because there are many fewer samples in each region than in the whole country, the results are subject to a lot of statistical uncertainty and patterns are rather hard to discern. There is no clear statistical evidence of differences in infection rates between different regions – such differences may exist, but, if they do, the survey hasn’t picked them up. The results for Wales show a similar proportion of the population being infected as in England, but this is only the second week that Welsh data have been published, and trends are not yet entirely clear. The other two sets of data, discussed next, are so far available only for England.
“The second set of figures gives estimates of the number of new infections (as would be measured by positive swab tests) each day in England. This number is different, and smaller, than the number of people who would provide a positive swab test if they were tested now, because that number would include people who had been infected for some time, not just those who have a new infection on the day in question. So, for example, the most recent estimate for the number of people in England who would test positive when tested for a current infection is 28,300, but the estimated number of new infections in the country per day is much smaller, at 3,800. That’s still too big a number to be ignored, of course – the virus hasn’t gone away – but you should see it in the context that there are 54.6 million people in the community population of England, and only about 7 in every hundred thousand of those people are getting newly infected each day, on average across the country.
“Because the number of new infections is considerably less than the total number of current infections, the estimates of new infections are subject to greater statistical uncertainty than the estimates of current infections. ONS report that, as for total infections, the rate of new infections fell until late June, and then increased slowly, but that trend ‘appears now to have levelled off’. I agree that there isn’t clear statistical evidence of a continuing increase – but because of the greater statistical uncertainty, I don’t think we can by any means conclude that there is definitely no increasing trend at present. There could possibly be such a trend, but there might well not be – we really can’t tell clearly from these data. And it definitely doesn’t look as if new infections could be falling.
“The third set of figures comes from antibody testing, and if someone has antibodies in their blood, that’s a sign that they were previously infected with SARS-CoV-2, possibly several months ago or possibly just a couple of weeks ago. The antibody testing results are less statistically precise than the swab (antigen) tests, because not all the people who provide swab tests in this survey are also asked to provide a blood sample for antibody tests. Also, only people aged 16 or over are included in the antibody testing – the swab tests include anyone aged two or more. This week’s estimate, but based on data from when the survey started in late April until 9 August, is that about 6% of the English community population would test positive for antibodies – that’s 2.8 million people, or about 1 in 16 of the population. But statistical uncertainty means that the data are consistent with anything between 1 in 13 and 1 in 19 of the population. Because of the lack of statistical precision, in previous weeks ONS did not provide any regional figures for antibody results. This week, for the first time, they do provide regional data. Unlike the position with current infections, there is statistically sound evidence of differences between some different regions in the percentages of people testing positive for antibodies. The percentages are higher, on this basis, in London than in Yorkshire and The Humber, the East Midlands, the South East and the South West of England. (Other differences between regions are not statistically significant.) This generally matches what is known from other sources about the course of the pandemic in different regions – remember that most past infections, that would lead to people having antibodies, would have taken place during the peak of the epidemic in March and April, and the number of infected people then was far higher than the number being infected in recent hotspots in places like Leicester or Blackburn. The regional results from this ONS survey are also generally similar to those from the Imperial College REACT antibody survey, published earlier this week. The REACT study was able to use rather larger samples of people than is so far the case for the ONS antibody survey, so it’s reassuring to find that they more or less match in terms of results.”
Dr Daniel Lawson, Lecturer in Statistical Science, School of Mathematics, University of Bristol said:
“SAGE appropriately warn they are not confident that R is currently below 1, for two reasons. Firstly, there is significant uncertainty in the data, and modelling this is very difficult. Secondly, the data are observed with a time lag, so we can only know R in the past with certainty. R is very close to 1, and these uncertainties combine to make it plausible – though not yet likely – that R is currently above 1.
“When the direct data is insufficient to address the question, statisticians look to additional data sources. The UK epidemic intervention easement is behind other European countries and we might use their progression to better estimate our own.
“These reports don’t use data from Europe, and properly modelling how our interventions will play out from the experience of other countries is very difficult. But infection is rising in many similar countries whose easement is more advanced than our own and this is cause for concern.”
Dr Yuliya Kyrychko, Reader in Mathematics, University of Sussex, said:
“Whereas the headline value for the range of R values has remained unchanged since last week, we should note that apart from the South East, for all other regions of England, the higher end of the R range is 1, or even 1.1 for the North West. Although there is uncertainty associated with estimating the value of R number, particularly when the numbers of new infections are low, the fact that the 7-day average of new daily infections has been steadily increasing since the middle of July suggests that it is very likely that the view of SAGE on R number not being below 1 is correct. What this means in practical terms is that the situation is pretty much on the precipice, and the it can develop very rapidly, which means that all possible caution should be taken by the public.”
Dr Konstantin Blyuss, Reader in Mathematics, University of Sussex, said:
“The latest data from the ONS suggests there were around 3,800 new daily infections in England in the week 3-9 August, while the results of the latest survey data indicate that only around 6% of the total population have had disease by the middle of July. This suggests that, on the one hand, the infection is still very prevalent in the population, and on the other hand, the majority of population is still susceptible to the infection, and a removal of any restriction currently in place carries with it a potential for growth in the number of infections. Since this is a directly transmitted infection, further local lockdowns may be required to reduce the spread and prevent a potentially very high second epidemic wave, that has already been seen in several European countries such as Romania, Bulgaria and Ukraine.”
* R number and growth rates: https://www.gov.uk/guidance/the-r-number-in-the-uk
** ONS Infection Survey: https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/coronaviruscovid19infectionsurveypilot/englandandwales14august2020
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 member of the SMC Advisory Committee, but my quote above is in my capacity as a professional statistician.”
None others received.