select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The Office for National Statistics (ONS), have released the latest data from their COVID-19 Infection Survey.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“Although the headline results from this week’s update of the ONS Coronavirus Infection Survey (CIS) came out, as is now usual, on Wednesday, there is some interesting and potentially important new information in today’s full weekly release.
“The main new information is on two topics. Today’s release updates the estimates of the rate at which people are becoming newly infected with the virus, and points to a possible difference in trends to what we’ve seen in the new confirmed cases on the Government dashboards at coronavirus.data.gov.uk. It also provides some details and comments on the BA.2 sub-variant of the virus, which is included under the Omicron name but differs in some respects.
“The headline information that came out on Wednesday is all about estimates of the numbers of people in the UK population who would test positive for the virus – that’s sometimes called the prevalence but a better term is the positivity rate. This week’s releases take the positivity estimates up to the week ending 29 January (last Saturday). Today’s ONS release adds more detail that was not in the headline release, particularly in giving modelled daily estimates for each of the UK countries and for individual regions and age groups in England, and there are some helpful charts for the regions and age groups. Overall, though, the new information doesn’t change the comments1 I made on Wednesday.
“Briefly, the decreases in infections, that were seen in the previous two weeks, were replaced by a pattern with some increases, particularly in Wales and Northern Ireland. Most of these changes were not large, but overall levels of infection do remain stubbornly high. The positivity levels in the English regions tended to decrease in the North and increase in the South, so that the differences between regions became smaller.
“In age groups, there were continuing increases in infection levels in children of school age and younger. If anything, though the trends aren’t clear, infection rates fell in most older age groups. The figure that ONS give for the week ending 29 January, for the infection rate in England in the youngest age group (age 2 to school year 6, so primary school age and younger), is the highest weekly figure for any age group in England since the CIS began there at the end of April 2020. ONS estimated that 13% of all the children in that age group would have tested positive on 26 January, their ‘reference day’ in that week. They say that this is 1 in 10 of the whole population in that age group in England – though ONS round those “1 in so many” figures to the nearest 5, and in fact the percentage corresponds to an infection rate slightly worse than 1 in 8. There’s a small bit of good news about that age group in the daily modelled figures, published today – they show the estimated rate beginning to fall between 26 January and the end of that week on the 29th, though it’s a small fall and that trend is too short-term to be much of a comfort yet.
“The main other estimates in the CIS releases are of the incidence rate – that is, of the numbers of newly infected people in a day or week. The positivity estimates refer to everyone who would test positive for the virus, whether they have just been infected or were infected a few days or even a few weeks ago. So, in many ways, the CIS incidence estimates are like the numbers of new confirmed cases on the Government dashboard. Those dashboard figures are prompter than the CIS estimates, partly because the main CIS prevalence estimates lag two weeks behind their positivity figures for technical reasons. But the dashboard case figures can be affected by changes in who is turning up for routine testing, by any problems with the availability of tests, and by the recent revision of the case numbers for England to include people who were previously infected on another occasion, the reinfections. So the CIS estimates can be more accurate because they are not affected by that kind of bias. A representative sample of people across the UK is tested for the CIS, regardless of whether they have symptoms. The CIS estimates have always included reinfections.
“The latest so-called ‘official’ prevalence estimates from the CIS go up to the week ending 15 January, but, unlike the positivity estimates, the modelled daily estimates of prevalence go on a little after that date, up to 19 January. For the last 7 days of estimates, 13 to 19 January, the total of the ONS estimates for the number of newly infected people is about 1.7 million. That’s lower than for the previous 7 days, when the total figure was about 1.9 million. (Both figures are subject to rather wide statistical margins of error, for technical reasons.)
“However, that overall decreases hides differences between the UK countries. In England and in Scotland, the total estimated (modelled) number of new cases for the week 13-19 January was lower than for the previous week, 6-12 January. But in Wales and in Northern Ireland the number in that most recent week was higher than the week before. In Wales and Northern Ireland, but also in in England as well, the trend of the daily modelled estimates for the latest week, 13-19 January, is upwards. (There’s no clear trend in Scotland really, though it’s downwards if anything.) The ONS bulletin says that the trend in the week ending 15 January was downwards in England, Wales, and Scotland, and unclear in Northern Ireland, but the latest modelled estimates going on after the end of that week show a slightly different picture.
“Because of the high levels of statistical uncertainty, none of these trends is entirely certain. But they don’t really match the trends in the dashboard new cases in England or Wales. That could matter. The dashboard cases for both those countries remained roughly level over the corresponding period, while the modelled CIS estimates rose a bit. Are the dashboard figures biased for some reason, or is the statistical uncertainty in the CIS estimates too great to say that there really is a discrepancy? I think we can’t tell yet for sure, but we should keep an eye on these patterns. Given the increases in some positivity estimates from the CIS, for the most recent week (ending 29 January), the incidence rate must have been increasing recently in some parts of the UK, but that’s not clearly showing up in the dashboard confirmed cases either.
“I’ll also mention, as I have before about earlier weeks, that the UK total of the CIS estimates of new infections for the week 13-19 January, about 1.7 million, is very much higher than the UK total of number of new cases on the Government dashboard for that week, which is just under 700,000. So again, more than half of the CIS estimated new infections don’t correspond to cases on the dashboard. There are many reasons why a new infection might not end up as a confirmed case on the dashboard. Probably the most important is that people who are infected but have no symptoms may not be tested, and if you aren’t tested you can’t be a confirmed case.
“On variants, the CIS figures again clearly show that the original Omicron variant, BA.1, remains the most common across all four UK countries. But ONS point out that “Our main variant analysis shows that in Northern Ireland, there was a notable increase in the percentage of people testing positive not compatible with Omicron BA.1 in the week ending 29 January 2022; there was also a small increase in England and early signs of an increase in Wales.” They say that the corresponding trend in Scotland was “uncertain”. ONS point out that people who test positive and are labelled as “not compatible with Omicron BA.1” are most likely to have the Omicron BA.2 variant.
“In England, Wales and Scotland, of the people whose positive test results were either “compatible with Omicron BA.1” or “not compatible with Omicron BA.1”, well under 10% were “not compatible with Omicron BA.1”. But in Northern Ireland, that percentage was about 30%, considerably higher. There would appear to be a lot of BA.2 there, compared to the position in the rest of the UK.
“Those figures all exclude smaller numbers of tests where the level of virus was too low to decide whether or not the results were compatible or not with BA.1. ONS refer to results being compatible with BA.1 or not, rather than saying they are BA.1 or BA.2, because the PCR tests that they use cannot tell for certain exactly which variant is involved. That can only be told by sequencing the genetic material – that is done for CIS samples where possible, but it isn’t always possible and takes longer than for the PCR results.
“I’m not a virologist and cannot comment on what these trends might mean for infectivity or disease severity.”
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.”