select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
The Office for National Statistics (ONS) have released the latest data from their COVID-19 Infection survey, and the government have released the newest estimates for the COVID-19 R value in the UK.
On both:
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“The weekly update on findings from the Office for National Statistics (ONS) Infection survey, as well as the updated estimates of the R number and the growth rate of new coronavirus infections from SAGE and the Government Office for Science, have both been published. My interpretation is that both of them indicate that the rate of new infections is approximately stable at the national level. It might be increasing slowly, or it could even be falling slowly. There is inevitable statistical uncertainty in these estimates. We just can’t be sure of the direction of change, if indeed there is a clear overall direction.
“Unlike many of the other sources of information on new cases, the ONS infection survey estimates aren’t affected by changes in the numbers of tests done as a result of people having symptoms. Those tested are a reasonably representative sample of the community population of England. (Wales and Northern Ireland are also now included, though no data have yet been published for Northern Ireland, and Welsh data have not been available for long enough to give any strong information on trends.) The infection survey indicated a slow increase in the number of infections from a low point at the end of June. However, ONS have been saying for several weeks now that this increase appears to have levelled off. I agree with that assessment, which they repeat again this week. (The latest report covers testing up to 25 August.) ONS, rightly, have to be cautious in what they say about the findings, because there is a considerable amount of statistical uncertainty in the figures. This is a large survey – almost 3,500 swab tests per day (on average) were carried out in the survey in the most recent fortnight, in England alone. But the current number of infections in England (and Wales) is low. Only 9 new infections were found, amongst the people in England tested for the survey, in the two weeks ending 25 August, so the rate of new infections just can’t be estimated all that precisely. Because of this uncertainty, ONS say that they cannot be confident that the rate of new cases in England is rising or falling, so they continue to report that it remains unchanged. I think that’s reasonable, but I’d add that if in fact it is rising or falling, it’s doing so pretty slowly.
“This position is also supported by the NHS Test and Trace data* that were published by the Department of Health and Social Care yesterday. Although the number of new cases reported for the most recent week (to 26 August) was higher than the previous week, the general trend for the past three weeks or so is pretty level, with no clear increase or decrease – and we have to take into account that, unlike the ONS Infection Survey numbers, Test and Trace numbers of new cases can be affected by the numbers of tests that are carried out and by the reasons that people come forward for testing in this programme.
“There are many reasons for uncertainty about the R number and the growth rate of new infections. That’s why the Government, rightly in my opinion, does not give a precise single estimate but instead gives a range of possible values into which it is likely that the true R number and the growth rate will fall. This week’s range for the R number, for the UK as a whole, is the same as last week and the week before, 0.9 to 1.1. If R is less than 1, the number of new cases is falling, and if R is greater than 1, it is increasing. The range includes values less and 1 and greater than 1, which indicates that it is not certain whether new cases are increasing or falling. The R number itself does not explicitly say how fast any increase or decrease might be, but the growth rate estimate does, and the range for that is -1% to +2% per day, indicating that the number of new cases could be falling by 1% every day, or rising by 2% every day, or anything in between. The true growth rate is more likely to be in the middle of that range that at either end. So, this week, the Government’s advisers might be saying that a small increasing rate is a bit more likely than a slow fall. Last week’s growth rate interval was from -2% to +1% per day, so the interval has moved up a little in the direction of increasing cases – but there is still uncertainty about the rate. It’s not even certain from this change in the range that the growth rate has definitely gone up, or definitely moved from the negative numbers (cases decreasing) to positive (cases increasing).
“It is reasonable to say that these numbers broadly match the ONS infection survey numbers. The detailed ONS figures that I described earlier are for England, not the whole UK. The Government publishes ranges of R numbers just for England, as well as for the UK as a whole. For England, the R number range is 0.9 to 1.0 – the upper end is a bit lower than last week when it was 1.1. But again, this change of the end of the range does not mean that the R number has definitely fallen for England, just that it might have fallen, and SAGE say that they cannot be confident that R in England is less than 1. The growth rate range for England is the same as last week, -2% to +1% per day. This is consistent with the ONS interpretation of the new infections from their infection survey – the rate might be changing slowly, either up or down, but it’s not moving much. The R and growth rate estimates might look as if they relate to the situation now, while the ONS infection survey results only go up to 25 August – but the R and growth rate figures are to a considerable extent based on data on things like hospitalisations and deaths, which lag behind data on new infections because it takes time, after a person is infected, until they might get ill enough to need to go to hospital.
“Very importantly, the report on the national R and growth rate estimates indicates that they are not necessarily very helpful in the current situation, where overall rates of infection across the country are low but there are peaks and outbreaks in certain places, and rates of new cases a long way below average in other places. The R number and growth rates are averages over the whole country. Regional rates are published too, but they tend to be subject to even more uncertainty than the national figures. In most regions the report says, rightly in my view, that the regional figures are “are insufficiently robust to inform policy decisions alone.” We need good local data on possible outbreaks in towns or smaller regions, and that comes from local monitoring of cases, hospitalisations and deaths, not from these broad national figures.”
Dr Yuliya Kyrychko, Reader in Mathematics, University of Sussex, said:
“Although the R number has seen very slight changes, it is important to combine it with the actual number of cases especially when the prevalence in the wider population is relatively low. There is almost no change on the rate of infections according to the ONS survey based on England and Wales data, but overall UK cases appear to be rising lately, with a lot of new positive cases registered in Scotland. At the moment, as with increased mobility due to reopening of schools, universities and overall return to office working, it is more important than ever that we have enough testing capacity to manage increased demand but also a robust track and trace system in place to be able to stop outbreaks before they become widespread.”
Dr Konstantin Blyuss, Reader in Mathematics, University of Sussex, said:
“It is reassuring to see that for several regions in England (Midlands, North East and Yorkshire, North West, and South East), the upper limit on the estimate of R number has reduced from 1.1 last week to 1.0 this week. Of course, this is accompanied by a caveat that by their very nature the estimates of R number are lagging behind the actual epidemiological situation by one to several weeks, and with small numbers of new cases, there is a higher level of uncertainty in these estimates.
“At the same time, ONS estimates suggest that only in England there were on average 2,000 new cases each day last week, indicating the infection is still quite prevalent in the population. Taken together with latest survey data indicating that only around 6% of the population have been exposed to the virus, this suggests that the potential for rapid new growth in the level of infection is still there, as is confirmed by a number of local outbreaks and some regions being declared as areas of concern. Combined with the fact that the seven-day average number of daily cases has been steadily increasing for the last several weeks, and the latest estimates of infection growth rate being between -1% and +2%, this clearly highlights the potential danger of growth in infections, and the importance of robust and efficient track-trace-isolate system to prevent local outbreaks from turning into something bigger that could require a countrywide lockdown.”
On ONS Infection Survey only:
Prof James Naismith FRS FMedSci, Director of the Rosalind Franklin Institute, and Professor of Structural Biology, University of Oxford, said:
“Conducting regular random surveys is the most informative way to determine both how widespread the virus is and its trajectory. It is therefore good news that within error there has been no significant uptick in the number of infections from July to August 25. The virus is still circulating but the opening up of the UK that began in June has not lead to surge in infections. We need to be alert going forward, the virus has not changed but it does appear that our behaviour has. The current strategy of continual gradual unlocking based on scientific advice has to this point worked. The science is uncertain and it is helpful for scientists make clear the current data does not support earlier doom laden predications. I am very pleasantly surprised that the rate has held constant, although I supported the gradual unlocking including schools I feared some of the steps, particularly opening pubs, might see a larger rise. Every new step brings risks, we have to be prepared to reverse a measure that results in a spike. We simply do not know enough to be sure of the outcome of each step in advance and false certainty helps no one. I remain confident that advances in medicine and science mean the proportion of infections that lead to death or serious illness will fall. I believe we will struggle to keep the numbers of infections constant as we head into the winter and there are issues with test, track and isolate that urgently need addressed. We have time to solve these problems but there is none to waste. If we can continue to contain the number of infections and keep the vulnerable virus free then we will avoid a second wave of serious illness.”
Prof Oliver Johnson, Professor of Information Theory, School of Mathematics, University of Bristol, said:
“Today’s ONS infection survey figures are very similar to last week. Indeed the long-term trend is broadly flat since the beginning of July, suggesting an R value very close to 1.
“This data may appear to contradict the recent increase in UK cases: this may partly be due to some of those cases being discovered by targeted testing in hotspots. Further, it is important to note this ONS survey covers only England and Wales: a significant proportion of the recent increase in cases has occurred in Scotland and Northern Ireland, and so would not be visible here.”
ONS Infection Survey
Government R Value Estimates
https://www.gov.uk/guidance/the-r-number-in-the-uk
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.